DESC:FIELD OF INVENTION
[001] The field of invention generally relates to sleep quality monitoring. More specifically, it relates to a system and a method for monitoring sleep quality through micro-vibrations.
BACKGROUND
[002] Poor sleep quality or sleep deprivation can impair cognitive function, reaction times, and alertness, increasing the risk of accidents and injuries, particularly while driving or operating machinery. Monitoring sleep quality is crucial for maintaining overall health, optimizing cognitive function, and regulating mood and emotional well-being.
[003] Conventional approaches to assessing sleep quality often rely on methods that are either invasive or impractical for regular use. For instance, traditional techniques such as polysomnography. In current methods, polysomnography (PSG) is considered a gold standard, but it is costly, inconvenient, and uncomfortable, requiring overnight stays in sleep labs. Similarly, other methods, such as electroencephalography (EEG), though effective in identifying sleep stages, involves placing electrodes on the scalp, making it impractical for home use and discomforting for users.
[004] Additionally, other existing methods provide valuable insights into physiological parameters but lack accuracy in assessing sleep stages. Actigraphy, while non-invasive, cannot accurately identify sleep stages, hence limiting its utility. Currently existing systems rely on proprietary algorithms, limiting interoperability and customization. Despite advancements in signal processing techniques, most of the current systems struggle to capture both objective physiological measurements and subjective user experiences effectively.
[005] Other existing systems have tried to address this problem. However, their scope was limited to either objective parameters such as vital signs or subjective sleep response parameters such as qualitative sleep assessments, without effectively integrating both aspects to provide a holistic evaluation of sleep quality.
[006] Thus, in order to overcome the shortcomings of the existing technologies, there is a need for a system and method for effectively capturing sleep quality of a user.
OBJECT OF INVENTION
[007] The principal object of this invention is to provide a system and method for monitoring sleep quality through micro-vibrations.
[008] A further object of the invention is to integrate objective physiological-based parameters and subjective sleep response parameters to offer a comprehensive assessment of sleep quality and associated health risks.
[009] Another object of the invention is to create personalized Artificial Intelligence (AI) or machine learning models that may be tailored to a user and can automate subjective sleep quality assessments based on user inputs during a human-in-the-loop (HITL) baseline period, thereby enhancing the efficiency and accuracy of sleep monitoring.
[0010] Another object of the invention is to include advanced signal processing techniques and machine learning techniques to detect specific clinical events during sleep, such as sleep apnoea, tachycardia, and snoring, for early intervention and prevention.
[0011] Another object of the invention is to provide a user-friendly interface that visualizes sleep trends and patterns, facilitating easy interpretation of sleep data and fostering informed decision-making regarding sleep habits and lifestyle adjustments.
[0012] Another object of the invention is to offer personalized interventions for improving sleep quality, such as targeted vibrations and temperature adjustments, based on real-time analysis of sleep metrics and user preferences.
[0013] Another object of the invention is to offer real-time, personalized sleep interventions, such as calibrated micro-vibrations or temperature modulation, based on ongoing analysis of sleep metrics and user preferences.
[0014] Another object of the invention is to establish a dynamic sleep score generator that evolves over time, combining machine learning models and longitudinal data to provide increasingly accurate, precise and personalized sleep insights.
[0015] Another object of the invention is to promote holistic well-being by not only monitoring sleep but also actively engaging users in improving their sleep habits and lifestyle choices through actionable recommendations and feedback.
SUMMARY
[0016] The aim of the present disclosure is to provide a system and method for monitoring sleep quality to enhance the accuracy and relevance of sleep assessments by combining objective physiological measurements and subjective user inputs along with the subjective perceived sleep parameters (generated by the artificial intelligence model). This is achieved through the implementation of machine learning models, micro-vibration analysis, and real-time feedback systems, as defined in the appended independent claims. Additional features and embodiments are set forth in the dependent claims. The aim of the present disclosure is achieved by a system and method for monitoring sleep quality as defined in the appended independent claims to which reference is made to. Advantageous features are set out in the appended dependent claims.
[0017] Throughout the description and claims of this specification, the words "comprise", "include", "have", and "contain" and variations of these words, for example "comprising" and "comprises", mean "including but not limited to", and do not exclude other components, items, integers or steps not explicitly disclosed also to be present. Moreover, the singular encompasses the plural unless the context otherwise requires. In particular, where the indefinite article is used, the specification is to be understood as contemplating plurality as well as singularity, unless the context requires otherwise.
BRIEF DESCRIPTION OF FIGURES
[0018] This invention is illustrated in the accompanying drawings, throughout which, like reference letters indicate corresponding parts in the various figures.
[0019] The embodiments herein will be better understood from the following description with reference to the drawings, in which:
Figure 1 depicts a system for monitoring sleep quality through micro-vibrations, in accordance with an embodiment of the present disclosure;
Figure 2A illustrates subcomponents of a sensor unit, in accordance with an embodiment of the present disclosure;
Figure 2B illustrates subcomponents of a sleep monitoring server, in accordance with an embodiment of the present disclosure;
Figure 2C illustrates subcomponents of a user device, in accordance with an embodiment of the present disclosure;
Figure 3A depicts subcomponents of a biomarker unit, in accordance with an embodiment of the present disclosure;
Figure 3B illustrates subcomponents of a quantitative assessor module, in accordance with an embodiment of the present disclosure;
Figure 3C depicts subcomponents of a qualitative estimator module, in accordance with an embodiment of the present disclosure;
Figure 4 depicts an exemplary embodiment for monitoring sleep quality through micro-vibrations, in accordance with an embodiment of the present disclosure;
Figure 5 illustrates a method for monitoring sleep quality through micro-vibrations, in accordance with an embodiment of the present disclosure; and
Figures 6A, 6B and 6C illustrate graphs that establish robust correlations between micro-vibration signal-derived features and validated subjective sleep quality measures, in accordance with an embodiment of the present disclosure.
DETAILED DESCRIPTION
[0020] The embodiments herein and the various features and advantageous details thereof are explained more fully with reference to the non-limiting embodiments that are illustrated in the accompanying drawings and/or detailed in the following description. Descriptions of well-known components and processing techniques are omitted so as to not unnecessarily obscure the embodiments herein. The examples used herein are intended merely to facilitate an understanding of ways in which the embodiments herein may be practiced and to further enable those of skill in the art to practice the embodiments herein. Accordingly, the examples should not be construed as limiting the scope of the embodiments herein.
[0021] The present invention discloses a system and method for monitoring sleep quality through micro-vibrations to provide a comprehensive assessment of a user's sleep experience. The system comprises a sensor unit placed in proximity to the user, such as beneath their mattress in a receiving unit that receives the user therein/thereon, to capture micro-vibrations associated with sleep to collect data to be fed into an AI model. By integrating objective sleep parameters and subjective user parameters, the system offers a comprehensive assessment of sleep quality and associated health risks. The system combines both objective physiological measurements (quantitative or measurable data) and subjective user experiences (qualitative data), which is reliable, non-invasive, accurate, comfortable, affordable and easy to use, facilitating widespread adoption and enhancing overall sleep health and well-being. Further, the system employs advanced signal processing techniques and machine learning models to enable detection of specific clinical events during sleep, while personalized Artificial Intelligence (AI) models that may be unique to a user and can automate subjective sleep quality assessments. One of the key advantages of the disclosed system is its ability to continue assessing sleep quality post-baseline period without requiring ongoing subjective input from the user; it intelligently extracts subjective sleep parameters from the micro-vibration data alone. This feature not only enhances user convenience but also ensures that the assessment remains consistent and reliable over time. By combining these elements, i.e., the objective sleep parameters and features, subjective sleep parameters, and advanced processing, the system provides a holistic view of sleep quality that empowers personalized insights, interventions, and actionable recommendations to improve sleep quality and overall well-being, fostering informed decision-making and proactive sleep management.
[0022] Figure 1 depicts a system 100 for monitoring sleep quality through micro- vibrations. The system 100 comprises a sensor unit 102, comprising at least one sensor, configured to measure micro-vibrations of a user; a user device 106 configured to provide a user interface to collect subjective sleep quality assessment data from the user, for a baseline period of sleep; and a at least one sleep monitoring server 104, communicably coupled to the sensor unit 102. The at least one sleep monitoring server 104 is configured to: receive the measured micro-vibrations and subjective sleep quality assessment data collected during the baseline period, extract objective sleep parameters and features from micro-vibration waveforms derived from the measured micro-vibrations, extract subjective sleep parameters from the collected subjective sleep quality assessment data, correlate the subjective sleep parameters with the features from the micro-vibration waveforms, for the baseline period, and generate at least one sleep score based on the objective sleep parameters and the subjective sleep parameters. Moreover, the at least one sleep monitoring server 104 is further configured to extract subjective sleep parameters from the features from the micro-vibration waveforms, after the baseline period, without collecting subsequent subjective sleep quality assessment data from the user.
[0023] Sleep quality is usually represented through objective and/or subjective parameters. Objective parameters are easily quantifiable which most wearables are also known to capture. These are parameters like vitals, sleep time, wake time, sleep stages, WASO, etc. However, the disclosed system also provides similar objective parameters without directly contacting the user, such as in case of wearable devices, and also maintains a history of this information to capture night-on-night variation for a given user. Beneficially, such variations over several sleeps along with the absolute values for a given sleep can contribute to the assessment of that sleep. The term "micro-vibrations" refers to minute oscillations or fluctuations in a physical medium, typically occurring at a frequency that may not be perceptible to human senses but can be quantified by specialized sensors.
[0024] In an embodiment, the micro-vibrations are generated by at least one of: cardiac activity, respiratory activity, body or body-part movement of the user. Herein, the terms "cardiac activity" and "respiratory activity" refer to the physiological processes associated with the heart's function (including heart rate and rhythm), and breathing (encompassing both inhalation and exhalation patterns), respectively. The term "body" refers to the entirety of the human organism, inclusive of all physiological systems and structures. The term "body-part movement" refers to the physical motion of specific anatomical segments, which may include limbs, torso, and head, as influenced by muscular and skeletal interactions.
[0025] The sensor unit 102 comprises at least one sensor and/or a processor placed in proximity to the user, such as embedded under a chair, a bed, mattress and/or pillow and the like, capable of detecting even subtle movements caused by cardiac and respiratory activity. In this regard, the sensor unit 102 is designed to capture these micro-vibrations by being placed in proximity to the user, allowing it to detect even minor fluctuations in movement. The sensor unit 102 is configured to identify sudden or slow signal deviations and amplitude fluctuations, which are indicative of body repositioning or external vibrations. Additionally, the sensor unit 102 is configured to differentiate between true physiological data and artifacts caused by sensor shifts or external disturbances. The sensor unit 102 is related to the at least one sleep monitoring server 104 as it provides the measured micro-vibrations, which are further processed to extract objective sleep parameters. Optionally, the sensor unit 102 is also related to the user device 106, as the user device 106 collects subjective sleep quality assessment data, which is later correlated with the data from the sensor unit 102. Moreover, the sensors that monitor heart rate (HR), respiration rate (RR), and movement data are integrated into bedding materials of the receiving unit, allowing for continuous data acquisition throughout the sleep cycle. By embedding vibration sensors under various surfaces like beds or pillows, of the receiving unit, the system effectively captures the micro-vibrations associated with the user's movements during sleep. This capability is particularly useful during sleep when physiological processes such as snoring, heart rate, and breathing patterns occur, providing valuable insights into the user's health and sleep quality. The accurate capture and analysis of micro-vibrations enhance the reliability of cardiac and respiratory data, leading to improved monitoring of the user's physiological state.
[0026] In an embodiment, the sensor unit 102, the at least one sleep monitoring server 104 and the user device 106 may be connected by at least one of wireless communication protocols such as Bluetooth, Wi-Fi, or cellular networks, enabling data transfer and interaction between the components of the sleep monitoring system.
[0027] In an embodiment, the sensor unit 102 comprises an array of piezoelectric sensors, selected from a group comprising BCG sensors and vibration sensors, configured to capture micro-vibrations of the user, wherein the sensor arrangement is arranged on a receiving unit configured to receive the user. In an embodiment, the system 100 comprises the sensor unit 102, implemented as at least one BCG (ballistocardiograph) sensor unit, at least one sleep monitoring server 104 and at least one user device 106. The BCG sensor unit comprises at least one BCG sensors, that are strategically placed to detect micro-vibrations caused by the user's physiological processes during sleep. The BCG sensors are capable of capturing subtle movements associated with cardiac and respiratory activities, which are indicative of various sleep parameters.
[0028] Moreover, the sensor unit 102 is configured to capture and analyse the ballistocardiograph (BCG) signals generated by the micro-vibrations of a user’s body during sleep, thereby providing one or more sleep parameters comprising the objective sleep parameters and features, that are used to assess sleep quality of the user. In an embodiment, the one or more sleep parameters herein are also referred to as micro vibration data.
[0029] In an embodiment, the objective sleep parameters comprise at least one of: heart rate (HR), respiration rate (RR), and movement data; and the features comprise at least one of:
- cardiac signals,
- sudden fluctuations in at least one of: heart rate or cardiac signal, in respiration rate or respiration signal, and body movements,
- sustained elevations in at least one of: heart rate, respiratory patterns, and body movements,
- sudden signal deviations, amplitude fluctuations, and non-cardiac waveform disturbances,
- ratio of standard deviation to mean of respiration rate, and
- portions where heartbeats slow down, respirations stabilize, and body movements are minimal.
[0030] Herein, the term "objective sleep parameters" refer to measurable physiological metrics that provide a quantitative basis for assessing sleep quality. Moreover, the term "features" refer to specific patterns or metrics extracted from the BCG signal that provide deeper insights into sleep quality. In an embodiment, the objective sleep parameters comprise at least one of heart rate, respiration rate, blood pressure, snoring, cardiac events, respiratory events, movement data, one or more sleep stages, and clinical event detection results. Typically, the features are derived using advanced signal processing and/or machine learning techniques.
[0031] The cardiac signals are derived from the ballistocardiogram (BCG) signal through advanced signal processing techniques that extract cardiac information by analysing successive J-J intervals representing individual heartbeats. The cardiac signals capture the subtle variations in cardiac rhythm during sleep, providing valuable insights into autonomic nervous system function and cardiovascular health that correlate with subjective sleep quality perceptions. Moreover, sudden fluctuations in heart rate, irregularities in respiration rate, and abrupt body movements during sleep are also derived from the BCG signal. Beneficially, quantifying these transient disturbances provides a comprehensive assessment of sleep fragmentation and micro-arousals that may not be consciously perceived by the user but significantly impact subjective sleep quality and next-day functioning. Furthermore, the ratio of the standard deviation to the mean of respiration rate is also derived from the BCG signal. The coefficient of variation metric provides critical insight into respiratory stability throughout the sleep period, with more stable breathing patterns typically associated with deeper, more restorative sleep and improved subjective sleep quality reports. Furthermore, portions of the BCG signals may be characterized by slowed heartbeats, stabilized respiration patterns, and minimal body movements, typically corresponding to periods of deep, undisturbed sleep that contribute significantly to the restorative function of sleep and correlate strongly with positive subjective sleep quality assessments. Furthermore, sustained elevations in heart rate, irregular respiratory patterns, and increased body movements are also derived from the BCG signal. It may be appreciated that the sustained elevations are different from sudden fluctuations as the sustained elevations focus on persistent rather than sudden physiological changes. These sustained alterations often indicate prolonged periods of lighter sleep or subtle stress responses that impact overall sleep architecture and subjective sleep quality. Furthermore, sudden signal deviations, amplitude fluctuations, and non-cardiac waveform disturbances may be identified and analysed from the BCG signal. These characteristics arise from body repositioning, external vibrations, or sensor shifts that affect heart and respiration waveform integrity. Through sophisticated artifact handling and signal processing, such alterations (comprising said deviations, fluctuations and disturbances) help distinguish true physiological data from movement artifacts, ensuring accurate cardiac and respiratory analysis that correlates with subjective sleep experiences.
[0032] Typically, cardiac signals are raw signals related to heart activity, extracted from the BCG data. The cardiac signals include the morphology of the cardiac waveform, which can be analysed to detect the user's heart rate by analysing successive J-J intervals (representing heartbeats). It may be appreciated that the cardiac signals may be used to derive Heart rate variability (HRV) and other cardiac events, to provide insights into autonomic nervous system activity during sleep. Similarly, the BCG signal may be analysed to detect respiratory cycles and calculate the respiration rate. This involves identifying patterns in the BCG signal that correspond to inhalation and exhalation. Similarly, body movements during sleep, including both subtle micro-movements and larger shifts in posture, may be detected from BCG signals. Beneficially, the movement data is critical for identifying restlessness, sleep fragmentation, and transitions between sleep stages. Notably, Sudden changes in these parameters indicate disturbances during sleep, such as stress or arousals, sleep apnoea events, respiratory disturbances, and so forth. Moreover, prolonged increase or sustained elevations in the one or more sleep parameters may indicate periods of poor sleep quality, such as during episodes of physiological stress, discomfort, breathing difficulties, restlessness, or sleep disorders like sleep apnoea. Moreover, external factors, such as sensor shifts, external vibrations, body repositioning, etc., may cause deviations (or noise or external artifacts) in the BCG signals. The ratio of standard deviation to mean of respiration rate quantifies the stability of the user's breathing during sleep. In this regard, a lower ratio indicates more stable respiration, which is associated with deeper, more restorative sleep, while, a higher ratio may indicate irregular breathing patterns, such as those seen in sleep apnoea. Moreover, portions where heartbeats slow down, respirations stabilize, and body movements are minimal indicate periods of deep sleep or NREM stages. Beneficially, combination of the objective parameters and the features enables a comprehensive assessment of sleep quality.
[0033] In an embodiment, the sensor unit 102 further comprises a secondary sensor, embedded in a head part of the receiving unit, configured to detect targeted eye movement, head motions, blinking and snoring. The term "secondary sensor" refers to a supplementary device or component that operates in conjunction with a primary sensor, namely the BCG sensor or vibration sensor, to capture additional data or enhance the accuracy of measurements related to an individual's physiological or environmental conditions. Herein, the secondary sensors are strategically arranged at the head part or head rest of the receiving using, such as a bed or chair, where the user is required to place his head during the analysis. The secondary sensor allows for close monitoring of signals originating from the head and face. Typically, eye movements, such as rapid and non-rapid during sleep, are accurately detected using the secondary sensor helps classify the user's sleep, which is critical for understanding sleep quality and patterns. The secondary sensor is further configured to detect blinking. Typically, blinking is more relevant during wakefulness, therefore, its monitoring can help in transitions between wakefulness and sleep. For instance, a decrease in blinking rate might indicate the onset of sleep. The secondary sensor also tracks head motions, which provides additional context for sleep analysis. For example, frequent head movements might indicate restlessness or disturbances during sleep, while minimal movement could suggest deeper sleep stages. The secondary sensor is further configured to detect snoring, thereby adding another layer of functionality to the system. Typically, snoring is often associated with sleep-disordered breathing, such as obstructive sleep apnoea (OSA), therefore, by monitoring snoring, the secondary sensor can provide insights into potential breathing irregularities during sleep, which may require further medical evaluation. Beneficially, the integration of multiple sensors allows for real-time data collection, ensuring that even the slightest movements during sleep are recorded. Such multi-sensor approach facilitates a comprehensive analysis of the user's physiological state, as the data from both sensors can be synchronized and processed together for improved accuracy.
[0034] In an embodiment, the at least one sleep monitoring server is further configured to classify sleep into sleep stages comprising REM, NREM1, NREM2, and NREM3, based on the objective sleep parameters and their frequency-domain characteristics. The term "sleep stage" refers to different categories of sleep, such as rapid eye movement (REM) and non-rapid eye movement (NREM). Typically, during REM sleep, the eyes move rapidly in various directions, and this activity is a hallmark of this sleep stage. By accurately detecting these movements, such as by using the secondary sensor, the system can determine when the user is in REM sleep, which is associated with dreaming and plays a vital role in cognitive and emotional health. However, NREM sleep is characterized by the absence of rapid eye movements. Notably, NREM is further subdivided into three stages (N1, N2, and N3) with N3 being the deepest sleep stage and N1 being the lightest sleep stage. Each stage has distinct physiological and neurological characteristics. For example, in NREM1 the body begins to relax, and the transition from wakefulness to sleep occurs. It is marked by slow eye movements and reduced muscle activity. In NREM2 is a deeper stage of sleep where heart rate slows, body temperature drops, and brain activity shows specific patterns like sleep spindles and K-complexes. NREM3, also known as slow-wave sleep (SWS), is characterized by high-amplitude, low-frequency brain waves (delta waves) and is critical for physical restoration and immune function. The secondary sensor's ability to detect the absence or presence of eye movements, along with other signals, helps classify the user's sleep into REM or NREM stages, which is critical for understanding sleep quality and patterns, diagnosing sleep disorders, and optimizing interventions. It may be appreciated that the secondary sensor is an optional extra sensor to improve accuracy. Even its absence, the system is able to classify sleep stages using the primary sensor. Moreover, the step of classifying sleep into the sleep stages may be done using HRV and movement based parameters, therefore, by using the primary sensors, and not just by detecting REM and/or NREM using the secondary sensors.
[0035] The term "frequency-domain characteristics" refers to the analysis of signals in terms of their frequency components. In this regard, the raw time-domain signals are transformed into their corresponding frequency domains using techniques like the Fourier Transform. In the frequency domain, different sleep stages exhibit distinct patterns. Beneficially, by analysing the frequency-domain characteristics of these signals, the processor and server can identify the unique patterns associated with each sleep stage.
[0036] Moreover, the system comprises a user device configured to provide a user interface to collect subjective sleep quality assessment data from the user, for a baseline period of sleep. Herein, the term "user device" refers to an electronic apparatus utilized by the user or an individual associated with the user (such as a caregiver, or nurse) to interact with the system. The system 100 may comprise as many user devices 106 as required by the users. The at least one user device 106 may comprise one or more of wearable device, mobile phones, PDA, smartphones, smart band, smart watch, laptop, computer, etc. The term "user interface" refers to the means by which the user or the individual engages with the system, facilitating input and output of information through visual, auditory, or tactile elements. The user interface may include touchscreens or voice recognition features, allowing users to easily provide feedback on their sleep experiences. The user interface facilitates the input of subjective sleep quality assessment data from the user. The term "subjective sleep quality assessment data" refers to information provided by the user regarding their personal perception of sleep quality, typically gathered through surveys or questionnaires. By utilizing prompts or questionnaires, the user device guides the user in evaluating various aspects of their sleep, such as duration, disturbances, and overall satisfaction. The system collects this data during a designated baseline sleep period, ensuring that the assessments reflect the user's typical sleep patterns.
[0037] The term "baseline period" refers to a predetermined duration of sleep that serves as a reference point for evaluating variations in sleep quality and associated metrics. In an embodiment, the baseline period may involve gathering information about user’s sleep habits, patterns, and subjective experiences over a set duration of time, such as several days or weeks. During the baseline period, the individual may be asked to provide subjective responses to questions or surveys related to their sleep quality, experiences, and any sleep-related issues they may encounter. These responses serve as a basis for understanding the individual's typical sleep behaviour and subjective perceptions of their sleep quality. The purpose of the baseline period is to establish a starting point or reference level against which future data and observations can be compared. Moreover, the baseline period is also used to personalise the AI model to automatically generate survey responses after the baseline period, without requiring the user to manually input responses to sleep assessment surveys. By establishing this baseline, it becomes possible to detect changes, trends, or deviations in sleep patterns and quality over time. This information is valuable for identifying potential sleep-related issues, evaluating the effectiveness of interventions or treatments, and providing personalized recommendations for optimizing sleep health and overall well-being. Beneficially, understanding individual sleep patterns and identifying potential issues that may affect overall health and well-being, enable the system to generate tailored recommendations for improving sleep, driven by the user's specific feedback.
[0038] In an embodiment, the user interface is further configured to:
receive contextual data, including room temperature, humidity, and noise levels;
receive biofeedback interventions, including vibration or temperature adjustments, based on detected disturbances;
display the sleep score and provide actionable sleep insights; and
provide a visual representation of sleep trends and patterns, including graphs, percentages, and color-coded scores.
[0039] In an embodiment, the contextual data may comprise information about the user's environment or lifestyle that may impact sleep quality. In this regard, the contextual data such as room or ambient temperature, temperature fluctuations may be important to assess as these parameters significantly impact sleep quality. For example, excessively high or low temperatures may disrupt sleep cycles. Similarly, high humidity can cause discomfort, while low humidity may lead to dryness in the airways. Similarly, background noise, such as snoring, traffic sounds, or other disturbances, may contribute to sleep fragmentation or poor sleep quality. For example, when a user logs the time, they consume caffeine before going to bed in the application, it provides contextual information about their sleep environment and habits. Beneficially, the contextual data provides a more holistic understanding of sleep patterns and factors affecting them, and allows for personalized recommendations.
[0040] Moreover, the term "biofeedback interventions" refers to real-time adjustments or therapeutic actions taken by the system to improve sleep quality. The biofeedback interventions are triggered based on the system's analysis of the user's physiological signals and environmental data. Optionally, the biofeedback interventions include personalized interventions for improving sleep quality, such as targeted vibrations and temperature adjustments, based on real-time analysis of sleep metrics and user preferences. In this regard, the system includes additional hardware features such as vibration motors and heating elements, that aid in relaxation and promoting better sleep practices.
[0041] Moreover, the user interface presents the at least one sleep score, which may be calculated by combining the objective parameters, the subjective parameters, and the contextual data. Moreover, the actionable sleep insights include, but are not limited to, recommendations for improving sleep habits (e.g., adjusting bedtime routines or reducing caffeine intake), alerts about potential sleep disturbances (e.g., signs of sleep apnoea or excessive restlessness), suggestions for optimizing the sleep environment (e.g., reducing noise or adjusting room temperature).
[0042] Furthermore, the user interface includes tools for visualizing sleep data over time, helping users track their progress and identify patterns. In an example, such visualizations include line or bar graphs showing trends in sleep metrics such as total sleep time, sleep stages (REM, NREM), and disturbances over days, weeks, or months. In another example, such visualizations include metrics like sleep efficiency (percentage of time spent asleep while in bed) or time spent in different sleep stages are displayed as percentages for easy understanding. In yet another example, such visualizations include color-coded sleep scores (e.g., green for good sleep, yellow for moderate sleep, and red for poor sleep) to provide an intuitive representation of sleep health. Beneficially, the visual representation allows users to quickly identify areas for improvement and track the effectiveness of interventions or lifestyle changes over time.
[0043] In an embodiment, the at least one user device 106 is configured to interact with the system 100 to provide a user-friendly user interface for viewing the at least one sleep score (received from at least one sleep score generator module 114, as discussed below). The user device 106 may facilitate user interaction by presenting sleep scores and actionable sleep insights, offering personalized recommendations and intervention strategies to each of the user, and enabling goal setting and progress tracking based on the provided feedback.
[0044] Furthermore, the system comprises the at least one sleep monitoring server, communicably coupled to the sensor unit. Herein, the term "at least one sleep monitoring server" 104 refers to a computational element that is operable to execute the software framework. Examples of the at least one sleep monitoring server 104 include, but are not limited to, a microprocessor, a microcontroller, a complex instruction set computing (CISC) microprocessor, a reduced instruction set (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, or any other type of processing circuit. Furthermore, the at least one sleep monitoring server may refer to one or more individual processors, processing devices and various elements associated with a processing device that may be shared by other processing devices. Additionally, one or more individual at least one sleep monitoring server, processing devices and elements are arranged in various architectures for responding to and processing the instructions that execute the software framework. The at least one sleep monitoring server 104 may comprise (or be associated with) at least one of a computer, a laptop, a tablet, external server, or a cloud server.
[0045] Further, the sensor unit 102 and the user device provide the micro-vibrations data (for extraction and assessment of one or more sleep parameters (comprising the objective sleep parameters and features) and the subjective sleep assessment data, respectively, to the at least one sleep monitoring server 104. Furthermore, the sleep processing unit 110 analyses and/ or processes the one or more sleep parameters to provide insights about sleep quality.
[0046] The at least one sleep monitoring server 104 is configured to extract objective sleep parameters and features from micro-vibration waveforms derived from the measured micro-vibrations, and extract subjective sleep parameters from the collected subjective sleep quality assessment data. The micro-vibration waveform refers to the signal generated by subtle, involuntary movements of the body during sleep, which are captured by highly sensitive sensors such as BCG sensors. Typically, the micro-vibration waveform represents the micro-vibrations (namely, the cardiac, respiratory, movement activities) as a time-series signal, which is analysed to extract physiological parameters like heart rate (HR), respiration rate (RR), and movement patterns.
[0047] In an embodiment, the at least one sleep monitoring server is further configured to filter noise and external artifacts from the measured micro-vibrations prior to extracting the objective sleep parameters therefrom. Notably, the micro-vibration waveforms are processed using advanced signal processing techniques to remove noise and artifacts, ensuring accurate extraction of sleep-related features and the objective sleep parameters as mentioned above. In this regard, the at least one sleep monitoring server preprocesses the signal data, filtering out noise and external artifacts, and transforming the signals into the frequency domain for analysis. Subsequently, the system classifies the sleep data into sleep stages (REM, NREM1, NREM2 or NREM3) based on the extracted features and their frequency-domain characteristics.
[0048] Herein, the subjective sleep parameter is a qualitative metric that reflects the user's perception of their sleep quality and its impact on their daily functioning. These parameters are typically gathered through self-reported surveys or questionnaires, namely the subjective sleep quality assessment data, and are influenced by the user's personal experiences and feelings. For example, perceived sleep quality (i.e., how well the user feels they slept), daytime alertness (i.e., the user's ability to stay awake and focused during the day), fatigue or energy levels (i.e., the user's sense of tiredness or vitality after waking up), emotional well-being (i.e., feelings of stress, confusion, or disorientation), and so on. Herein, during the baseline period, the user provides, via the user interface, responses to subjective sleep quality assessment (surveys or questionnaires), namely, the subjective sleep quality assessment data.
[0049] Moreover, the at least one sleep monitoring server 104 is configured to correlate the subjective sleep parameters with the features from the micro-vibration waveforms, for the baseline period, and generate at least one sleep score based on the objective sleep parameters and the subjective sleep parameters. For example, a user reporting daytime fatigue may have corresponding patterns of sleep fragmentation or reduced deep sleep in the micro-vibration data. Similarly, a user reporting high perceived stress may show sustained elevations in heart rate or irregular respiratory patterns during sleep. In this regard, the subjective sleep parameters are subsequently mapped to the features derived from the micro-vibration waveforms, using machine learning models to automate their estimation and provide a holistic sleep quality assessment.
[0050] Herein, the term "sleep score" refers to a numerical representation that summarizes an individual's sleep quality based on the analysis of collected data and established criteria.
[0051] Further, the at least one sleep monitoring server 104 comprises a sleep processing unit 110. The sleep processing unit 110 further comprises a sleep assessment module 112 and a sleep score generator module 114. The sleep assessment module 112 provides sleep quality assessment by combining both objective sleep parameters, subjective sleep response parameters and contextual data. The sleep quality assessment data is further transmitted to the sleep score generator module 114. In an embodiment, the at least one sleep score generator module 114 is configured to receive the assessment and generate at least one sleep score based on objective sleep parameters, subjective sleep response parameters, and contextual data. Thereafter, at least one sleep score and actionable sleep insights are generated by score generator module 114 based on the assessments and the sleep score is shared with the user device 106. Furthermore, the user device 106 facilitates user interaction by presenting the sleep scores and the actionable sleep insights, offering personalized recommendations and intervention strategies to each of the user, and enabling goal setting and progress tracking based on the provided feedback.
[0052] In an embodiment, the at least one sleep monitoring server is further configured to detect sleep-time clinical events, including sleep apnoea, tachycardia, snoring, restlessness, and sleep fragmentation, sleep disruptions, sleep stages, based on sudden fluctuations in the objective sleep parameters. Throughout the present disclosure, the term "sleep apnoea" refers to a sleep disorder characterized by repeated interruptions in breathing during sleep, often resulting in reduced oxygen levels. The term "tachycardia" refers to an abnormally rapid heart rate, which may occur during periods of sleep disturbance. The term "snoring" refers to the sound produced during sleep due to the vibration of respiratory structures, often associated with obstructed airflow. The term "restlessness" refers to a state of unease or inability to remain still during sleep, which may lead to frequent awakenings. The term "sleep fragmentation" refers to the disruption of sleep continuity, resulting in multiple awakenings and a reduction in overall sleep quality. The term "sleep disruptions" refers to any disturbances that interrupt the normal sleep cycle, affecting the duration and quality of sleep. The term "sleep stages" refers to the various phases of sleep, including REM (rapid eye movement) and non-REM stages, each characterized by distinct physiological and neurological activity. Beneficially, by continuously monitoring these parameters, the system can identify deviations that indicate potential clinical events during sleep. In this regard, for example, the at least one sleep monitoring server monitors respiration rate (RR) and respiratory patterns derived from the BCG signal. Notably, sudden drops in RR or prolonged pauses in the respiratory signal indicate apnoea events. Additionally, compensatory changes, such as increased heart rate (HR) or abrupt body movements, often follow apnoea episodes. Moreover, sudden, sustained elevations in heart rate (HR) beyond normal sleep thresholds are flagged as tachycardia events. Snoring is indicated by low-frequency vibrations in the BCG signal which correspond to snoring sounds. Furthermore, sudden, irregular spikes in the movement signal or frequent posture changes are flagged as restlessness. Additionally, sudden fluctuations in HR, RR, and movement data indicate arousals or brief awakenings, and frequent transitions between sleep stages (e.g., from deep sleep to light sleep), namely sleep fragmentation. Moreover, sudden signal deviations or amplitude fluctuations in the BCG waveform caused by external factors (e.g., noise, vibrations) or internal factors (e.g., body repositioning, irregular breathing) may result in sleep disruptions.
[0053] In an embodiment, the system further comprises a database unit 108 configured to store: objective sleep parameters, features, subjective sleep parameters, at least one sleep score, user inputs, historical sleep data for short and long-term trend analysis, actionable sleep insights resulting from the short and long-term trend analysis. The database unit is a storage system that organizes and maintains all the data collected and processed by the system. In an embodiment, the at least one database unit 108 is configured to store the one or more sleep parameters. In an embodiment, the database unit 108 may be positioned within the at least one sleep monitoring server 104. In another embodiment, the database unit 108 may be positioned outside the at least one sleep monitoring server 104 as an external database which is linked to the database unit 108. In an embodiment, the database unit 108 may be stored in an external server, a cloud server or in the user device 106.
[0054] Optionally, the one or more sleep parameters and the subjective sleep assessment data are stored inside the database unit 108 for further analysis and processing. The user inputs may include demographic data (e.g., age, gender, weight) and lifestyle/activity data (e.g., caffeine consumption, exercise routines, work schedules). Beneficially, user inputs help contextualize the sleep data and personalize the analysis. Moreover, the historical sleep data refers to the collection of past sleep-related information stored in the database unit 108, wherein the historical sleep data includes both objective and subjective sleep parameters, as well as any contextual information related to the user's sleep environment and lifestyle. It may be appreciated that the disclosed system maintains the historical sleep data to capture night-on-night variation for a given user. This variation over several sleeps along with the absolute values for a given sleep can contribute to the assessment of that sleep. The term "short-term trend analysis" refers to the examination of sleep data over a brief timeframe to identify immediate fluctuations or changes in sleep quality. The term "long-term trend analysis" refers to the evaluation of sleep data over an extended duration to discern persistent patterns and shifts in sleep quality and associated health implications. The historical sleep data also allows to automate some of the PSQI and similar subjective sleep quality assessment survey data which require users to average out sleep habits over the past week, month, etc.
[0055] Furthermore, the at least one sleep monitoring server 104 is further configured to extract subjective sleep parameters from the features from the micro-vibration waveforms, after the baseline period, without collecting subsequent subjective sleep quality assessment data from the user. It may be appreciated that during the baseline period, the system collects (manually by the user) both the subjective responses and the micro-vibration waveform features and objective sleep parameters. During this period, the sleep assessment module in the at least one sleep monitoring server maps the subjective responses to the corresponding features extracted from the micro-vibration waveforms, using a machine learning model or algorithms.
[0056] In an embodiment, the at least one sleep monitoring server is configured to employ a machine learning model, trained during a baseline period using the subjective sleep quality assessment data. The term "machine learning model" refers to a computational framework that utilizes algorithms to analyse and interpret data patterns, enabling the system to make predictions or decisions based on input data without explicit programming for each specific task. In this regard, initially, the machine learning model is a base model, namely, a generic 'pretrained' model, that is trained on several subjects' data already available. It may be appreciated that the pretrained model is not tuned to a specific user. The algorithms are trained on extensive datasets to identify patterns and correlations in the data, enabling the extraction of key features like heart rate variability and respiratory rate. During the baseline period for a given user, the pretrained model will be available for use in prepopulating survey responses. As a generic model, the pretrained model generates survey responses. Optionally, the user can correct the pretrained model's responses, as needed. In this regard, the algorithm processes the subjective sleep parameters collected during a given baseline period to identify patterns and correlations between subjective assessments and features from the micro-vibrational waveforms. In this regard, the system may further employ natural language processing techniques to interpret the qualitative feedback provided by users, converting subjective assessments into quantifiable metrics. The system is designed to recognize patterns in the responses over successive sleep cycles, allowing it to adjust its automation strategies accordingly.
[0057] However, after the baseline period, the system no longer requires the user to manually provide subjective responses. In other words, after the baseline period, the generic pretrained model gets retrained, using transfer learning concept, on the specific responses of the given user from the baseline period. It may be appreciated that the term "training" as used here refers to an ongoing process of training and/or retraining based on incoming user inputs. The term "transfer learning" refers to a machine learning technique where an ML model developed for a first task is reused as the starting point for an ML model on a second task, by leveraging the knowledge gained from training on a large dataset to improve the learning efficiency and performance on a new, related task, especially when the new task has limited data. In this regard, the retrained machine learning model uses the features extracted from the micro-vibration waveforms to automatically estimate the subjective sleep parameters or each successive sleep session, using the trained machine learning model. In this regard, the at least one sleep monitoring server 104 extracts features from the micro-vibration waveforms, such as cardiac signal features (e.g., heart rate variability); sudden fluctuations in heart rate, respiration rate, or movement (indicating disturbances); respiratory stability (ratio of standard deviation to mean respiration rate); periods of minimal movement, stable respiration, and slowed heart rate (indicating deep sleep); sustained elevations in heart rate or irregular respiratory patterns (indicating stress or arousals); signal deviations caused by artifacts (e.g., body repositioning or external vibrations), and so on. The at least one sleep monitoring server 104 subsequently correlates the subjective responses (e.g., perceived stress, daytime sleepiness) with the extracted features, and the machine learning model learns to predict subjective sleep parameters based on the features. This results in a personalised ML model for that particular user. After the baseline period, the personalised ML model is configured to prepopulate survey responses for the given user with improved accuracy. The user may still continue to correct the personalised ML model responses, as needed. Any such inputs from the user will lead to further tuning and improvement of the personalised ML model, thus retraining via transfer learning.
[0058] For subsequent sleeps, the at least one sleep monitoring server 104 predicts subjective responses based on the micro-vibration data and historical sleep data. By employing machine learning model, the system refines its predictive capabilities, allowing it to adapt to individual user profiles to facilitate real-time analysis and interpretation of physiological data. The incorporation of diverse data points enhances the model's accuracy in assessing sleep quality over time, ensuring that it remains responsive to changes in user behaviour and sleep patterns. Additionally, the system automates the estimation of subjective sleep parameters by leveraging both current and historical data.
[0059] Optionally, the system allows the user to review the predicted responses and make adjustments if necessary. Herein, the term "adjustment" refers to correcting the responses generated by the pretrained ML models. In this regard, the user may enter correct assessment in case there is any discrepancy, during generation of the personalized model as discussed above. Moreover, the system refines its predictions over time by comparing them with any optional user feedback, by employing transfer learning. For example, if the user occasionally reviews and adjusts the predicted responses, the AI model incorporates this feedback to improve its accuracy. Beneficially, having a human-in-the-loop (HITL) architecture enhances the accuracy of the system.
[0060] Optionally, the ML model integrates unsupervised and supervised learning techniques. The ML model employs peak clustering algorithms for identifying unique patterns in heart and breathing rates captured through the BCG signals. The peak clustering algorithms analyse the temporal distribution and morphological characteristics of signal peaks, by grouping similar patterns to distinguish between normal cardiac cycles, respiratory events, and potential anomalies. By applying unsupervised clustering methods to these peaks, the system can identify recurring patterns that correspond to specific physiological states, enabling accurate classification of sleep stages and detection of sleep disturbances without requiring manual annotation or calibration.
[0061] Moreover, the system employs parametric statistical models for analysing BCG waveform or micro-vibrational waveform morphology, extracting critical information about cardiac and respiratory function during sleep. Typically, the parametric statistical models mathematically characterize the shape, amplitude, duration, and variability of individual waves within the BCG signal or micro-vibration waveforms, creating statistical representations of normal and abnormal physiological patterns. By fitting these parametric statistical models to incoming data, the system can quantify subtle deviations from expected waveform characteristics, providing insights into cardiovascular health, respiratory effort, and autonomic nervous system activity that correlate with subjective sleep quality measures and potential health concerns.
[0062] Furthermore, the system employs deep learning networks tailored to understand complex physiological signals. The analysis leverages both time-domain and frequency-domain approaches, as mentioned above, emphasizing spectral features to discern harmonics and map relative frequency strengths and artifacts.
[0063] Furthermore, the system employs specialized pretrained supervised models capable of distinguishing high-quality BCG signals from noise-contaminated data. The pretrained supervised models are trained on spectral features within frequency bands specifically relevant to patterns emerging from heartbeats and respiration cycles. By analysing the spectral composition of incoming signals and comparing them against learned representations of clean versus noisy data, the system can automatically identify and filter out environmental interference, sensor artifacts, and non-physiological vibrations. This sophisticated noise detection approach ensures that only high-fidelity physiological data contributes to sleep quality assessments, maintaining accuracy even in challenging home environments.
[0064] Furthermore, supervised and unsupervised machine learning models may be combined to enable precise detection and classification of movements during sleep by analysing piezoelectric sensor values to distinguish between BCG signals corresponding to small movements, large positional changes, and periods of no movement. The supervised component leverages labelled training data to recognize known movement patterns, while the unsupervised component identifies novel movement signatures without prior examples. Beneficially, such comprehensive movement detection contributes to accurate sleep staging, as well as provides valuable insights into restlessness during sleep, helping quantify sleep fragmentation and its impact on overall sleep quality and next-day functioning.
[0065] In an embodiment, the at least one sleep monitoring server is configured to automate a response to a subjective sleep quality assessment survey for successive sleep cycles based on the features from the micro-vibration waveforms. Subjective sleep quality assessment surveys are tools used to measure a user's perceived sleep quality and its impact on their daily functioning. Typically, the subjective sleep quality assessment surveys may be in the form of questionnaires comprising questions related to daytime alertness and energy levels, difficulty staying awake during activities (e.g., driving, eating, or socializing), feelings of disorientation, confusion, or fatigue, effort required to maintain enthusiasm or focus, and so on. Examples of the subjective sleep quality assessment surveys include the Pittsburgh Sleep Quality Index (PSQI), Epworth Sleepiness Scale (ESS), and Perceived Stress Scale (PSS), to which the user submits answer manually. Alternatively, additional sleep assessment tools may include, but are not limited to, Insomnia Severity Index (ISI), Berlin Questionnaire, Stanford Sleepiness Scale (SSS), Sleep Diary, STOP-Bang Questionnaire.
[0066] Typically, Pittsburgh Sleep Quality Index (PSQI) is a self-rated questionnaire that assesses sleep quality and disturbances over a one-month interval. It consists of 19 individual items that generate seven component scores related to subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medication, and daytime dysfunction. These component scores are summed to produce a global PSQI score ranging from 0-21, with higher scores indicating poorer sleep quality. A global score greater than 5 is considered indicative of poor sleep quality. Epworth Sleepiness Scale (ESS) is a simple, self-administered questionnaire that measures a user's general level of daytime sleepiness. It asks users to rate their likelihood of falling asleep in eight different everyday situations on a scale from 0 (would never doze) to 3 (high chance of dozing). The total score ranges from 0-24, with scores above 10 suggesting significant daytime sleepiness that may require medical attention. The ESS is widely used to screen for sleep disorders characterized by excessive daytime sleepiness, such as obstructive sleep apnoea and narcolepsy. Perceived Stress Scale (PSS) is a psychological instrument that measures the degree to which situations in a user's life are appraised as stressful. The scale consists of 10 items that ask about feelings and thoughts during the last month, with responses ranging from 0 (never) to 4 (very often). Scores range from 0-40, with higher scores indicating higher perceived stress. The PSS measures how unpredictable, uncontrollable, and overloaded respondents find their lives, factors that have been shown to correlate with sleep quality and various health outcomes.
[0067] Moreover, Insomnia Severity Index (ISI) is a brief screening measure of insomnia and assesses the nature, severity, and impact of insomnia. The 7 items included in ISI are rated on a 5-point scale, with the total score indicating the severity of insomnia. Berlin Questionnaire is designed to identify patients at risk for obstructive sleep apnoea, it categorizes patients based on their risk level using questions related to snoring behaviours, wake-time sleepiness or fatigue, and the presence of obesity or hypertension. Stanford Sleepiness Scale (SSS) measures the user's level of sleepiness at a particular moment. It uses a 7-point scale, where a higher score indicates greater sleepiness. Sleep Diary is a self-reported log where users record details about their sleep habits over a period of time. This can include bedtime, wake time, total sleep time, quality of sleep, and any night-time awakenings. There are multiple formats associated with this which could be generated as a report. STOP-Bang Questionnaire is a screening tool for obstructive sleep apnoea that includes questions about snoring, tiredness, observed apnoea, high blood pressure, BMI, age, neck circumference, and gender.
[0068] Additionally, Mindful Attention Awareness Scale (MAAS) may be used to measure dispositional mindfulness. The MAAS is a 15-item psychological assessment tool designed to measure the capacity to maintain awareness and attention to present-moment experiences in daily life. Users rate statements about everyday experiences on a scale from 1 (almost always) to 6 (almost never), with higher scores indicating greater mindfulness. The MAAS evaluates a user's tendency to function on "automatic pilot" versus maintaining conscious awareness of thoughts, emotions, sensations, and behaviours. This scale is widely used in mindfulness research and has been shown to correlate with psychological well-being, cognitive functioning, and sleep quality.
[0069] Beneficially, through these established correlations, the system transforms traditionally subjective sleep quality assessments into objective, quantifiable metrics derived from non-contact BCG sensor measurements. This approach enables automated, continuous monitoring of both physiological sleep parameters and their impact on subjective well-being, providing a comprehensive sleep quality assessment without requiring daily user input after an initial calibration period.
[0070] In an example, a user who reports moderate daytime fatigue and difficulty concentrating during the baseline period, the at least one sleep monitoring server correlates the user responses with features such as increased movement during sleep (indicating restlessness), frequent transitions between sleep stages (indicating sleep fragmentation), and elevated heart rate during certain periods of sleep (indicating stress or arousals). For future sleep cycles, the at least one sleep monitoring server may use these correlations to predict the user's subjective responses based on the extracted features. If the at least one sleep monitoring server detects similar patterns of restlessness and fragmentation, it predicts moderate fatigue and difficulty concentrating. Conversely, if the features indicate stable sleep with minimal disturbances, the at least one sleep monitoring server predicts improved daytime alertness and focus.
[0071] Beneficially, automating a response to a subjective sleep quality assessment survey for successive sleep cycles, enhances user convenience, improves the accuracy of subjective assessments, and provides a comprehensive understanding of sleep quality over time.
[0072] In an embodiment, the at least one sleep monitoring server is configured to generate alert notifications for the user, based on sleep-time clinical events by analysing the historical sleep data. In this regard, the at least one sleep monitoring server uses machine learning models and advanced signal processing techniques to analyse the historical sleep data and extracts features such as HRV, respiratory stability, movement patterns, sleep stage transitions, signal deviations, etc., to identify patterns and trends in the extracted features from the historical sleep data. Optionally, the patterns and trends may include repeated occurrences of clinical events (e.g., frequent apnoea or tachycardia episodes), gradual changes in sleep quality metrics (e.g., declining deep sleep percentage or increasing restlessness), correlations between contextual data (e.g., caffeine intake) and sleep disturbances, and so forth. The at least one sleep monitoring server subsequently applies predefined thresholds or machine learning models to detect abnormalities, such as excessive HR beyond a predefined threshold, multiple episodes of apnoea within a predefined interval of time, excessive restlessness compared to historical trends, and so forth. When the at least one sleep monitoring server detects a sleep-time clinical event, it generates the alert notification for the user, in real-time or near-real-time. The alert notification is provided via the user interface. Optionally, the alert notification may be implemented as a visual alert, an audio alert, a tactile alert, a text alert, or a combination thereof. Optionally, the alert notification may comprise a description of the detected sleep-time clinical event (such as "High heart rate detected during sleep"), supporting data or visualizations data (such as graphs showing HR trends over time), recommendations for further action, namely, actionable sleep insights (such as "Consider consulting a healthcare provider" or "Reduce caffeine intake before bed"), and so on. Beneficially, the alert notifications allow for personalized, actionable sleep insights that help users identify and address potential health risks, improving their overall sleep quality and well-being.
[0073] In an embodiment, the at least one sleep monitoring server is configured to dynamically update the at least one sleep score by incorporating incoming data from successive sleep sessions and refining the machine learning model. It may be appreciated that the sleep score is not static, it evolves over time as the system collects more data (namely, the micro-vibrations, subjective sleep quality assessment data, contextual data, detected clinical events) from successive sleep sessions. This dynamic nature ensures that the score reflects both short-term variations (e.g., a single night of poor sleep) and long-term trends (e.g., gradual improvement or decline in sleep quality). In this regard, the at least one sleep monitoring server updates the new data to the user's existing sleep history, creating a more comprehensive record of their sleep patterns. Subsequently, at least one sleep monitoring server compares the new data with historical data to detect changes in sleep quality, such as increased restlessness or improved deep sleep duration. The at least one sleep monitoring server uses the machine learning (ML) models to analyse the collected data and generate the sleep score, and in the process, the ML models are continuously refined as new data becomes available. Specifically, the ML model learns from the new data to improve its ability to predict subjective sleep parameters (e.g., perceived stress, daytime fatigue) based on features extracted from the micro-vibration waveforms and detect clinical events with greater accuracy by identifying subtle patterns in the data. Moreover, the ML model becomes increasingly personalized to the user as it learns from their unique sleep patterns and responses. For example, if the user consistently experiences high perceived stress after nights with frequent sleep fragmentation, the model adjusts its predictions to account for this correlation. Beneficially, dynamically updating the sleep score ensures that the sleep score becomes increasingly accurate, personalized, and reflective of the user's evolving sleep patterns and health conditions.
[0074] Moreover, the system allows the user to provide feedback and incorporates feedback from the user (if provided) to correct any inaccuracies in the predicted subjective parameters or sleep score. For example, if the user edits their perceived stress level after reviewing the system's prediction, this feedback is used to fine-tune the model.
[0075] In an embodiment, the at least one sleep score is represented as a weighted index combining objective sleep parameters and the subjective sleep parameters. A weighted index is a mathematical representation where different components such as objective and subjective sleep parameters) are assigned specific weights based on their relative importance or contribution to the overall score. In this regard, the objective sleep parameters are given a weight based on their reliability and accuracy in assessing sleep quality, and the subjective sleep parameters are also weighted to reflect their importance in capturing the user's perception of sleep quality, which may not always align with objective metrics. In an example, the objective parameters like HR, RR, and sleep stages may be given higher weights because they are directly measurable and reliable, while the subjective parameters like perceived stress or daytime fatigue may be given slightly lower weights but are still critical for capturing the user's experience. The final sleep score is calculated by combining these weighted components into a single, unified metric, namely the weighted index. Notably, the weighted index is determined through predefined algorithms, machine learning models, or expert-defined rules. Optionally, the system dynamically adjusts the weighting of different features (e.g., HR, RR, movement data) based on their relevance to the user's sleep quality. For example, if respiratory stability is found to have a stronger correlation with the user's perceived sleep quality than movement data, the model increases the weight assigned to F3 in the sleep score calculation. Beneficially, such weighted index ensures that the sleep score provides a comprehensive and holistic assessment of sleep quality by combining measurable data with the user's personal experiences and perceptions.
[0076] In an embodiment, the weighted index is normalized to a standard scale (such as 1-10, a numerical value, a percentage value, a colour-coded visual, and so forth), for an easy user interpretation.
[0077] In an embodiment, the at least one sleep monitoring server is further configured to employ an identification module to identify the user, based on a set of physical features of the user. The identification module in the sleep monitoring server uses a combination of physical features derived from micro-vibration data, such as cardiac signals, respiratory patterns, and movement data, to accurately identify the user. Optionally, other features, such as body weight distribution (i.e., pressure exerted on the mattress by the user's body) and/or sleep posture (i.e., preferred sleeping positions and posture changes during sleep) can influence the BCG signal and serve as an identifying feature for the user. In this regard, the identification module employs machine learning (ML) algorithms to analyse the physical features and identify the user. Notably, during the baseline period, the system trains the identification module using the extracted features, to recognize the unique patterns associated with each user. For each new sleep session, the system compares the incoming data with the stored profiles in the historic data module. The identification module uses the trained model to match the features of the current session with the closest stored profile, identifying the user. In the process, the system continuously refines the identification model by incorporating new data from successive sleep sessions, improving its accuracy over time. Beneficially, the identification module embedded in the system ensures that the system can accurately associate sleep data with the correct user, even in multi-user environments, such as shared beds or households.
[0078] For example, User A and User B share the same bed. During the baseline period, the system collects micro-vibration data for both users and creates unique profiles based on their cardiac, respiratory, and movement features. User A has a resting heart rate of 60 bpm, a respiration rate of 14 breaths per minute, and minimal movement during sleep. User B has a resting heart rate of 75 bpm, a respiration rate of 18 breaths per minute, and frequent posture changes. For each subsequent sleep session, the system captures the micro-vibration data and compares it with the stored profiles. If the data matches User A's profile, the system attributes the sleep session to User A and generates personalized insights and recommendations. If User B's profile is detected, the system attributes the session to User B.
[0079] Figure 2A illustrates subcomponents of a sensor unit 102, in accordance with an embodiment of the present disclosure. Herein, the sensor unit 102 is implemented as a BCG sensor unit. In an embodiment, the BCG sensor unit comprises at least one BCG sensor 204 and at least one biomarker unit 206. The BCG sensor 204 comprises at least one sensor placed in proximity to the user. The BCG sensor 204 may be configured to detect even subtle movements caused by cardiac and respiratory activity, called as detected BCG sensor data. The at least one sensor may comprise at least one of vibration sensors and BCG sensors. The one or more vibration sensors may be embedded under a seat, a bed, a chair, a mattress and/or a pillow of the receiving unit to capture subtle movements or micro-vibrations caused by cardiac and respiratory activity during sleep. These micro-vibrations / BCG sensor data are indicative of one or more of physiological processes, comprising, snoring, blood pressure, cardiac events, respiratory events, body temperature, heart rate, breathing patterns, and body movements.
[0080] In an embodiment, the at least one BCG sensor 204 may also comprise a sensor that may be placed under the pillow (not shown in figure) to pick up actual rapid eye movement to improve rapid eye movement (REM) and non-rapid eye movement (N-REM) staging accuracy.
[0081] Further, the BCG sensor 204 transmits the detected BCG sensor data to the biomarker unit 206. The biomarker unit 206 comprises at least one processor. The biomarker unit 206 is configured to collect and pre-process the BCG sensor data received from the BCG sensor 204 for further analysis.
[0082] Figure 2B illustrates subcomponents of the sleep monitoring server 104, in accordance with an embodiment of the present disclosure. The at least one sleep monitoring server comprises at least one database unit 108, at least one sleep processing unit 110, a memory module 218 and a communication module 220. In an embodiment, the at least one database unit 108 comprises a historic data module 214 and a user response data module 216. The historic data module 214 comprises a record of the one or more sleep parameters, comprising at least one of heart rate, respiration rate, movement, blood pressure, snoring, respiration events, cardiac events, and sleep stages, extracted from past sleep sessions. The user response data module 216 comprises qualitative sleep parameter responses provided by the user during at least one baseline period, along with at least one of demographic and activity data. In an embodiment, the demographic data may be information about the user’s characteristics, such as age, gender, height, weight and the like. For example, the user’s age is recorded as 35 years old in a sleep monitoring application profile. These factors can influence sleep patterns and overall health, so consideration of demographic data allows for more personalized sleep assessments and recommendations. Further, activity data may be information about the user’s daily activities and lifestyle habits. The demographic data may comprise data from wearable devices or smartphone applications that track physical activity levels, exercise routines, daily routines, work schedules, screen time, caffeine or alcohol consumption, and other relevant behaviours. By incorporating activity data into the system 100, it becomes possible to analyse how lifestyle factors impact sleep quality and to provide tailored recommendations for improving sleep habits.
[0083] In an embodiment, the at least one sleep processing unit 110 is configured to analyse and/ or process the one or more sleep parameters received from the at least one sensor unit 102 (implemented as BCG sensor unit) to provide insights about sleep quality. The at least one sleep processing unit comprises at least one sleep assessment module 112 and at least one sleep score generator module 114. In an embodiment, the at least one sleep assessment module 112 is configured to analyse the data collected by the at least one BCG sensor unit 102 and stored in the at least one database unit 108. In an embodiment, the at least one sleep assessment module 112 comprises a quantitative assessor module 208 and a qualitative estimator module 210.
[0084] The quantitative assessor module 208 is configured to analyse objective sleep parameters to assess sleep quality based on predefined machine learning and artificial intelligence (AI) algorithms and models. The quantitative assessor module 208 utilizes features such as heart rate, snoring, blood pressure, respiration rate, movement patterns, and sleep stages to generate quantitative sleep quality assessments. The quantitative assessor module 208 quantifies objective parameters of the current sleep parameters, by fetching the sleep parameters for the historic sleeps and using the parameters and biomarkers for the current sleep.
[0085] In an embodiment, the qualitative estimator module 210 is configured to map subjective sleep response parameters provided by the user during the baseline period to corresponding features extracted from the micro-vibration waveform. The qualitative estimator module 210 trains a machine learning model to learn from user responses collected for a sleep during the baseline period to map micro vibration data to subjective sleep response parameters for that sleep. Further, after the completion of the baseline period, the qualitative estimator module 210 automates subjective sleep response parameters estimation of the user, by using the micro-vibrations data from that sleep and historic sleep data for the user.
[0086] In an embodiment, the sleep score generator module 212 is configured to integrate objective and subjective sleep quality assessments generated by the quantitative assessor module 208 and the qualitative estimator module 210 respectively.
[0087] The sleep score generator module 212 applies predefined algorithms to weigh importance of the one or more features and inputs and combine them to generate a comprehensive sleep score. The predefined algorithms may be at least one of artificial intelligence algorithms and machine learning algorithms.
[0088] In an embodiment, the at least one sleep score generator module 114 may also generate a comprehensive sleep score based on objective sleep parameters, subjective sleep response parameters, and contextual data.
[0089] In an embodiment, the memory module 218 is configured to store generated sleep scores for future reference and analysis, data and one or more instructions to be executed.
[0090] In an embodiment, the communication module 220 facilitates communication between one or more components of the system, as well as with external devices or user interfaces, enabling data exchange and feedback dissemination. The communication module 220 may comprise wired and wireless communication, including but not limited to, GPS, GSM, LAN, Wi-fi compatibility, Bluetooth low energy as well as NFC. The wireless communication may further comprise one or more of Bluetooth (registered trademark), ZigBee (registered trademark), a short-range wireless communication such as UWB, a medium-range wireless communication such as Wi-Fi (registered trademark) or a long-range wireless communication such as 3G/4G or WiMAX (registered trademark), according to the usage environment.
[0091] Figure 2C illustrates subcomponents of a user device 106, in accordance with an embodiment of the present disclosure. In an embodiment, the at least user device 106 may facilitate user interaction by presenting sleep scores and insights, offering personalized recommendations and intervention strategies to each of the user, and enabling goal setting and progress tracking based on the provided feedback. In an embodiment, the at least one user device 106 comprises a sleep application 222, a processing unit 224, a memory unit 226, and a communication unit 226.
[0092] The sleep application 222 incorporates a Large Language Model (LLM) with three primary functions. Firstly, notifying the user about unusual sleep patterns or trends. Further, engaging in conversations with the user to inform and discuss sleep assessments. Thereafter, facilitating discussions around sleep quality improvement.
[0093] The processing unit 224 within the at least one user device 106 executes one or more algorithms and models for data analysis and interpretation. In an embodiment, the processing unit 224 may comprise one or more microprocessors, circuits, and other hardware configured for processing.
[0094] In an embodiment, the memory unit 226 of the at least one user device 106 comprises one or more volatile and non-volatile memory components which are capable of storing data and instructions to be executed.
[0095] In an embodiment, the communication unit 226 of the at least one user device 106 may comprise wired and wireless communication, including but not limited to, GPS, GSM, LAN, Wi-fi compatibility, Bluetooth low energy as well as NFC. The wireless communication may further comprise one or more of Bluetooth (registered trademark), ZigBee (registered trademark), a short-range wireless communication such as UWB, a medium-range wireless communication such as Wi-Fi (registered trademark) or a long-range wireless communication such as 3G/4G or WiMAX (registered trademark), according to the usage environment.
[0096] Figure 3A depicts subcomponents of the biomarker unit 206. The biomarker unit 206 comprises a data collection unit 302 and data processing unit 304. In an embodiment, the data collection unit 302 is configured to gather physiological signals during sleep from the BCG sensor 204. The data collection unit 302 collects data related to the user's sleep patterns and physiological parameters, comprising heart rate, snoring, blood pressure, respiration rate, movement patterns, cardiac events, respiration events and other relevant parameters. The identification and categorization of movement data/ patterns and artifacts related to restlessness and posture changes enhances the accuracy of sleep quality assessment of the system. In an embodiment, the data processing unit 304 comprises biomarker feature extraction unit 314, biomarker AI (Artificial intelligence) model 310, time-domain analysis unit 312 and frequency domain analysis unit 312. The data processing unit 304 is configured to preprocess raw data collected by the data collection unit 302, extract relevant features, and analyse the data to derive biomarkers indicative of sleep quality.
[0097] The noise detected through BCG sensor data is removed through techniques such as filtering, signal averaging, and wavelet denoising. Filtering methods, such low-pass, and high-pass filters, target specific frequency bands to eliminate unwanted noise. Signal averaging combines multiple measurements to reduce random noise, while wavelet denoising decomposes the signal into frequency bands and thresholds the coefficients to suppress noise. Additionally, adaptive noise cancellation and artifact rejection techniques may be utilized to further enhance the accuracy of the processed data, ensuring reliable analysis of sleep-related physiological signals.
[0098] The biomarkers may be computed using signal processing and statistical methods. Further, the biomarkers may also use supervised and/or unsupervised machine learning model in addition to signal processing techniques. The Biomarker AI model 310 within the data processing unit 304 aids in interpreting complex patterns and extracting meaningful insights from the data. The data processing unit 304 may use clustering algorithm to identify unique patterns in heart rate and breathing rate data. By clustering similar peaks in the data, the system can detect patterns indicative of different physiological states or events.
[0099] Furthermore, the data processing unit 304 may use statistical models to analyse morphology of waves presents in the physiological signals. By fitting statistical models to the waveform data, the system can extract parameters that characterize the shape, duration, and amplitude of the waves, providing insights into the underlying physiological processes.
[00100] Further, the data processing unit 304 may use deep learning algorithms to understand complex physiological signals and extract meaningful features. These deep learning algorithms are trained on large datasets to learn hierarchical representations of the data, enabling the system to automatically identify patterns and correlations that may not be apparent through traditional methods.
[00101] In an embodiment, the biomarker feature extraction unit 314 comprises machine learning algorithms and methods for extracting key features from the raw sensor data. These key features may comprise heart rate variability, respiratory rate, blood pressure, cardiac events, respiratory events, REM, movement patterns, and other relevant metrics. The machine learning algorithms extract detailed features from the data, such as frequency, amplitude, and temporal patterns. These features serve as valuable indicators of sleep quality, overall physiological status, and the presence of potential clinical events like sleep apnoea, tachycardia, snoring, and breathlessness.
[00102] In an embodiment, the biomarker AI model 310 comprises machine learning algorithms trained to recognize patterns and correlations within the extracted features. The biomarker AI model 310 is configured to learn from the data and improve its predictive capabilities over time.
[00103] In an embodiment, the time-domain analysis unit 308 comprises algorithms and techniques for analysing signals in the time domain, such as calculating heart rate variability, respiratory rate variability, and other temporal characteristics of the physiological signals.
[00104] In an embodiment, the frequency domain analysis unit 312 comprises algorithms and methods for analysing signals in the frequency domain, such as performing Fourier transforms to identify frequency components associated with different physiological phenomena. The frequency domain analysis unit 312 is configured to extract spectral features and characterize the frequency distribution of the physiological signals during sleep.
[00105] Figure 3B illustrates subcomponents of a quantitative assessor module 208. The quantitative assessor module 208 comprises a historic database unit 314, a biomarker data unit 316 and an assessor feature unit 318. The quantitative assessor module 208 quantifies objective parameters of the current sleep parameters, by fetching the sleep parameters for the historic sleeps from the historic database unit 314 and using the biomarkers for the current sleep that it gets from the biomarker data unit 316.
[00106] The historic database unit 314 comprises a repository of past sleep data, comprising physiological features and behavioural features such as heart rate, blood pressure, cardiac events, respiratory events, respiration rate, movement patterns and sleep stages, extracted from previous sleep sessions. The historic database unit 314 is configured to store and retrieve this data for analysis by the quantitative assessor module 208. The physiological features may comprise measurements such as heart rate, respiratory rate, heart rate variability. The behavioural features may comprise sleep patterns, sleep architecture (distribution of sleep stages), movement patterns, sleep position, subjective sleep quality, and observations of daytime functioning.
[00107] The biomarker data unit 316 provides biomarkers for the current sleep. The biomarkers comprise at least one of heart rate, respiratory events, cardiac events, body temperature, blood pressure, respiratory rate and the like.
[00108] The assessor feature unit 318 comprises artificial algorithms and methods for extracting key features from the objective sleep parameters obtained from the historic database unit 314. These features may comprise heart rate variability, blood pressure, respiratory events, and cardiac events, movement patterns, and other relevant parameters. The assessor feature unit 318 is configured to preprocess the data and extract features that contribute to the assessment of sleep quality.
[00109] Figure 3C depicts subcomponents of the qualitative estimator module 210. The qualitative estimator module 210 comprises a questionnaire module 320, a user response module 322, and an estimator AI (Artificial Intelligence) model 324. The qualitative estimator module 210 is configured to automate subjective sleep quality assessments based on user responses during a baseline period and micro-vibrations data.
[00110] The questionnaire module 320 comprises predefined questions designed to collect subjective qualitative sleep parameter responses from the user during the baseline period. The questionnaire model is configured to present questions to the user and record their responses for analysis by the estimator AI model 324. For example, the questions may comprise “How alert and focused do you feel during the day?", "Do you experience daytime sleepiness or fatigue?", "Are you able to concentrate and perform daily tasks effectively?”.
[00111] The user response module 322 facilitates the collection and processing of user responses to the questionnaire model. The user response module 322 presents questions to the user, records their responses, and provides feedback to the estimator AI model 324 for training and refinement.
[00112] In an embodiment, the user may use the system during sleep and fill out the questionnaire after waking up. This is repeated for a certain number of days or weeks which is called the baseline period.
[00113] The estimator AI model 324 is trained using the micro-vibrations data and the user responses data collected during the baseline period. Further, the estimator AI model 324 learns to map the subjective sleep parameter responses provided by the user for a sleep to corresponding features extracted from the sleep data or micro vibration data for that sleep. The estimator AI model 324 is configured to learn from user responses and adjust its mapping algorithms over time to improve accuracy. For any sleep after the baseline period, this estimator AI model 324 automatically generates the subjective responses for that sleep on the user's behalf. The estimator AI model 324 does so by using micro-vibrations data from that sleep and historic sleep data for the user.
[00114] Figure 4 depicts an exemplary embodiment for monitoring sleep quality through micro-vibrations. Firstly, the user 402 prepares for sleep by lying down on the mattress 404. The mattress 404 is equipped with the at least one BCG sensor unit comprising one or more sensors strategically placed underneath its surface. As the user 402 settles onto the mattress 404, the one or more sensors are activated. These one or more sensors are designed to detect and capture even the slightest movements generated by the user's 402 body during sleep. Once activated, as the user settles in for sleep, these one or more sensors begin to collect data in real-time, capturing subtle micro-vibrations generated by the user’s body during sleep, including cardiac and respiratory activity, as well as body shifts.
[00115] This data, along with subjective sleep response parameters such as user reported sleep experiences and feelings, is then transmitted to a data processing unit, typically located within the system’s hardware or connected to a nearby at least one user device. Optionally, the data processing unit is near the user, wherein the processing may happen on edge near the user or on the cloud. In the data processing unit, the collected signals undergo thorough analysis. They are time-synchronized and processed to extract detailed features that serve as indicators of sleep quality, physiological status, and potential clinical events.
[00116] As the user 402 wakes up the morning, comprehensive sleep quality reports are generated, offering insights into both objective metrics such as sleep duration, heart rate variability, and any disturbances, as well as subjective experiences reported by the user 402. It tracks changes in micro-vibrations, identifies sleep stages, and detects any irregularities or disturbances that may occur during sleep. If any significant events or abnormalities are detected, the system may provide real time feedback or alerts to the user 402 or caregiver, notifying them of potential issues such as sleep apnoea, snoring, or restless sleep.
[00117] It may be appreciated that various embodiments and variants disclosed above, with respect to the aforementioned system, apply mutatis mutandis to the method as well.
[00118] In a preferred embodiment, the method comprises:
measuring micro-vibrations of a user using a non-contact sensor unit;
collecting subjective sleep quality assessment data from the user using a user device configured to provide a user interface, for a baseline period of sleep;
extracting objective sleep parameters and features from micro-vibration waveforms derived from the micro-vibrations;
extracting subjective sleep parameters from the collected subjective sleep quality assessment data;
correlating the subjective sleep parameters with the features from the micro-vibration waveforms, for the baseline period; and
generating at least one sleep score based on the objective sleep parameters and the subjective sleep parameters,
wherein the method further comprises extracting subjective sleep parameters from the features from the micro-vibration waveforms, after the baseline period, without collecting subsequent subjective sleep quality assessment data from the user.
[00119] Herein, the non-contact sensor unit refers to an arrangement of a plurality of sensors that gather comprehensive sleep and health data without directly contacting with the user. By placing an array of sensitive vibration or BCG sensors under the mattress, the system captures the subtle movements caused by cardiac, respiratory and body movement activities. Moreover, these signals are time-synchronized and processed to extract detailed features that serve as indicators of sleep quality, physiological status, and potential clinical events such as sleep apnoea, tachycardia, snoring, and breathlessness.
[00120] Figure 5 illustrates a method 500 for monitoring sleep quality through micro vibrations.
[00121] The method 500 begins with utilizing at least one sensor unit 102 to detect and analyse one or more sleep parameters of a user’s body during sleep, as depicted at step 502. Subsequently, the method 500 discloses transmitting the detected and analysed one or more sleep parameters data to at least one sleep monitoring server 104 for storage and further analysis, as depicted at step 504.
[00122] Additionally, the method 500 discloses providing a sleep quality assessment by combining both objective sleep parameters, subjective sleep response parameters and contextual data using a sleep assessment module 112, thereby providing sleep assessment data, as depicted at step 506.
[00123] Furthermore, the method 500 discloses transmitting the sleep assessment data by the sleep assessment module 112 to a sleep score generator module 114, wherein the sleep assessment module 114 generates a sleep score using the sleep assessment data, as depicted at step 508.
[00124] Thereafter, the method 500 discloses communicating the generated sleep score to at least one user device 106 by the at least sleep monitoring server 104, allowing a user interaction by presenting sleep scores and insights, offering personalized recommendations and intervention strategies to each of the user, and enabling goal setting and progress tracking based on the provided feedback, as depicted at step 510.
[00125] In an embodiment, the method further comprises displaying via the user interface the at least one sleep score and associated actionable sleep insights to a user.
[00126] In an embodiment, the method further comprises employing a machine learning model, trained during a baseline period using the subjective sleep quality assessment data.
[00127] In an embodiment, the method further comprises storing historical sleep data in a database unit; and using the stored data to refine the machine learning model for personalized sleep quality monitoring.
[00128] In an embodiment, the method further comprises automating a response to a subjective sleep quality assessment survey, for successive sleep cycles based on the features from the micro-vibration waveforms.
[00129] In an embodiment, the method further comprises generating alert notifications for the user, based on sleep-time clinical events by analysing the historical sleep data.
[00130] Figures 6A, 6B and 6C illustrate graphs that establish robust correlations between micro-vibrations derived features and validated subjective sleep quality measures.
[00131] As shown in Figure 6A, for the Perceived Stress Scale (PSS), a weighted combination of features,
F1 – cardiac signals;
F2 – sudden fluctuations;
F3 – ratio of SD to Mean of RR;
F4 – portions of BCG signals with slow heartbeats, stabilized respiration, and minimal body movements; and
F5 – sustained elevations,
demonstrates clear differentiation between low (0-13), moderate (14-26), and high (27-40) perceived stress levels. This correlation enables the system to objectively quantify subjective stress perceptions that traditionally require self-reporting, providing insights into how sleep quality impacts next-day stress resilience.
[00132] As shown in Figure 6B, for the Epworth Sleepiness Scale (ESS), which measures daytime sleepiness, a weighted algorithm incorporating features
F1 – cardiac signals;
F2 – sudden fluctuations;
F4 – portions of BCG signals with slow heartbeats, stabilized respiration, and minimal body movements; and
F5 – sustained elevations,
accurately distinguishes between five clinically significant categories: lower normal (0-5), higher normal (6-10), mild excessive (11-12), moderate excessive (13-15), and severe excessive (16-24) daytime sleepiness. This correlation allows the system to predict likely daytime functioning impairments based solely on overnight BCG measurements, eliminating the need for daily subjective assessments.
[00133] As shown in Figure 6C, for Mindful Attention Awareness Scale (MAAS), the system further correlates all six features (F1-F6),
F1 – cardiac signals;
F2 – sudden fluctuations;
F3 – ratio of SD to Mean of RR;
F4 – portions of BCG signals with slow heartbeats, stabilized respiration, and minimal body movements;
F5 – sustained elevations; and
F6 – sudden signal deviations, amplitude fluctuations and non-cardiac waveform disturbances,
demonstrating clear differentiation between low (1-2), moderate (3-4), and high (5-6) mindfulness states. This correlation is particularly significant as it connects objective sleep measurements with cognitive and attentional capacities the following day, providing a comprehensive view of how sleep quality impacts higher-order mental functioning.
[00134] The advantages of the current invention comprise its non-invasive nature, as the micro-vibration sensors placed beneath the mattress allow for seamless and comfortable sleep monitoring without the need for wearable devices or invasive procedures.
[00135] An additional advantage is that the system provides both objective and subjective sleep quality assessments, combining physiological data with user reported experiences to offer a comprehensive understanding of sleep patterns and behaviours.
[00136] An additional advantage is the system's real-time feedback capability, enabling users to receive immediate alerts or notifications if any abnormalities or disturbances in sleep patterns are detected, allowing for timely interventions or adjustments.
[00137] An additional advantage is the system's ability to generate personalized sleep insights and trends over time, empowering users to make informed decisions and modifications to their sleep habits for improved overall well-being.
[00138] An additional advantage is the system's scalability, with the potential for future enhancements and updates to further improve sleep monitoring accuracy and functionality, ensuring long-term relevance and effectiveness.
[00139] An additional advantage is the potential for the system to contribute to advancing sleep research and understanding, by providing valuable data insights and facilitating collaborative studies in the field of sleep science and medicine.
[00140] Specifically, the disclosed system provides various benefits over the conventional methods. For example, it provides benefit over the PSQI and similar survey in automation and objectivity. In this regard, the system is implemented as a human-in-the-loop system where the user's response to PSQI survey is captures over a few days, referred to as a baseline period of sleep. Such user response is subsequently used to automate future responses accurately for the days that follow. This will be more objective and thus more accurate. It will also capture long term daily data to be able to assess changes early.
[00141] Moreover, the system provides benefit over polysomnography and is compliant for home use and comfort, and does not need an expert. Moreover, the system provides benefit over some wearables owing to its contactless nature.
[00142] Furthermore, the system provides benefit over other BCG system and audio systems owing to its improved accuracy over pure movement, snoring and HRV algorithms. A plurality of additional features like HR, RR, BP, movements, and sleep staging estimations (REM and NREM) may be determined using the BCG sensor data.
[00143] Furthermore, overall, the disclosed system is more comfortable, more accurate, able to assess changes in the long term, and creates automation.
[00144] Applications of the current invention comprise sleep monitoring homes, healthcare facilities, research laboratories and the like.
[00145] The foregoing description of the specific embodiments will so fully reveal the general nature of the embodiments herein that others can, by applying current knowledge, readily modify and/or adapt for various applications such specific embodiments without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. Therefore, while the embodiments herein have been described in terms of preferred embodiments, those skilled in the art will recognize that the embodiments herein can be practiced with modification within the scope of the embodiments as described here. ,CLAIMS:Claims
I/We Claim:

1. A system for monitoring sleep quality, the system comprising:
a sensor unit, comprising at least one sensor, configured to measure micro-vibrations of a user;
a user device configured to provide a user interface to collect subjective sleep quality assessment data from the user, for a baseline period of sleep; and
at least one sleep monitoring server, communicably coupled to the sensor unit, configured to:
receive the measured micro-vibrations and subjective sleep quality assessment data collected during the baseline period,
extract objective sleep parameters and features from micro-vibration waveforms derived from the measured micro-vibrations,
extract subjective sleep parameters from the collected subjective sleep quality assessment data,
correlate the subjective sleep parameters with the features from the micro-vibration waveforms, for the baseline period, and
generate at least one sleep score based on the objective sleep parameters and the subjective sleep parameters,
wherein the at least one sleep monitoring server is further configured to extract subjective sleep parameters from the features from the micro-vibration waveforms, after the baseline period, without collecting subsequent subjective sleep quality assessment data from the user.

2. The system as claimed in claim 1, wherein the micro-vibrations are generated by at least one of: cardiac activity, respiratory activity, body or body-part movement of the user.

3. The system as claimed in claim 1, wherein
the objective sleep parameters comprise at least one of: heart rate (HR), respiration rate (RR), and movement data; and
the features comprise at least one of:
- cardiac signals,
- sudden fluctuations in at least one of: heart rate, cardiac signal, in respiration rate, respiration signal, and body movements,
- sustained elevations in at least one of: heart rate, respiratory patterns, and body movements,
- sudden signal deviations, amplitude fluctuations, and non-cardiac waveform disturbances,
- ratio of standard deviation to mean of respiration rate, and
- portions where heartbeats slow down, respirations stabilize, and body movements are minimal.

4. The system as claimed in claim 1, wherein the at least one sleep monitoring server is further configured to detect sleep-time clinical events, including but not limited to sleep apnoea, tachycardia, snoring, restlessness, and sleep fragmentation, sleep disruptions, sleep stages, based on sudden fluctuations in the objective sleep parameters.

5. The system as claimed in claim 1, wherein the user interface is further configured to:
receive contextual data, including room temperature, humidity, and noise levels;
receive biofeedback interventions, including vibration or temperature adjustments, based on detected disturbances;
display the at least one sleep score and provide actionable sleep insights; and
provide a visual representation of sleep trends and patterns, including graphs, percentages, and color-coded scores.

6. The system as claimed in claim 1, further comprising a database unit configured to store: objective sleep parameters, features, subjective sleep parameters, at least one sleep score, user inputs, historical sleep data for short and long-term trend analysis, actionable sleep insights resulting from the short and long-term trend analysis.

7. The system as claimed in claim 6, wherein the at least one sleep monitoring server is configured to employ a machine learning model, trained during a baseline period using the subjective sleep quality assessment data.

8. The system as claimed in claim 6, wherein the at least one sleep monitoring server is configured to automate a response to a subjective sleep quality assessment survey for successive sleep cycles based on the features from the micro-vibration waveforms.

9. The system as claimed in claim 1, wherein the sensor unit comprises an array of piezoelectric sensors, selected from a group comprising BCG sensors and vibration sensors, configured to capture micro-vibrations of the user, wherein the sensor arrangement is arranged on a receiving unit configured to receive the user.

10. The system as claimed in claim 9, wherein the sensor unit further comprises a secondary sensor, embedded in a head part of the receiving unit, configured to detect targeted eye movement for determining rapid eye movement (REM) and non-rapid eye movement (NREM) along with head motions, blinking and snoring.

11. The system as claimed in claim 1, wherein the at least one sleep monitoring server is further configured to classify sleep into sleep stages comprising REM, NREM1, NREM2, and NREM3, based on the objective sleep parameters and their frequency-domain characteristics.

12. The system as claimed in claim 7, wherein the at least one sleep monitoring server is configured to generate alert notifications for the user, based on sleep-time clinical events by analysing the historical sleep data.

13. The system as claimed in claim 1, wherein the at least one sleep monitoring server is configured to dynamically update the at least one sleep score by incorporating incoming data from successive sleep sessions and refining the machine learning model.

14. The system as claimed in claim 1, wherein the at least one sleep score is represented as a weighted index combining objective sleep parameters and the subjective sleep parameters.

15. The system as claimed in claim 1, wherein the at least one sleep monitoring server is further configured to filter noise and external artifacts from the measured micro-vibrations prior to extracting the objective sleep parameters therefrom.

16. The system as claimed in claim 1, wherein the at least one sleep monitoring server is further configured to employ an identification module to identify the user, based on a set of physical features of the user.

17. A method for monitoring sleep quality, the method comprising:
measuring micro-vibrations of a user using a non-contact sensor unit;
collecting subjective sleep quality assessment data from a user using a user device configured to provide a user interface, for a baseline period of sleep;
extracting objective sleep parameters and features from micro-vibration waveforms derived from the micro-vibrations;
extracting subjective sleep parameters from the collected subjective sleep quality assessment data;
correlating the subjective sleep parameters with the features from the micro-vibration waveforms, for the baseline period; and
generating at least one sleep score based on the objective sleep parameters and the subjective sleep parameters,
wherein the method further comprises extracting subjective sleep parameters from the features from the micro-vibration waveforms, after the baseline period, without collecting subsequent subjective sleep quality assessment data from the user.

18. The method as claimed in claim 17, further comprising displaying via the user interface the at least one sleep score and associated actionable sleep insights to a user.

19. The method as claimed in claim 17, further comprising employing a machine learning model, trained during a baseline period using the subjective sleep quality assessment data.

20. The method as claimed in claim 19, further comprising:
storing historical sleep data in a database unit; and
using the stored data to refine the machine learning model for personalized sleep quality monitoring.

21. The method as claimed in claim 17, further comprising automating a response to a subjective sleep quality assessment survey, for successive sleep cycles based on the features from the micro-vibration waveforms.

22. The method as claimed in claim 17, further comprising generating alert notifications for the user, based on sleep-time clinical events by analysing the historical sleep data.