1. A system (100) for predicting an electronic device addiction of a user, the system (100) comprising: a camera (102), adapted to capture a live video stream of a face of a user, wherein the camera (102) is built into the electronic device; an input interface (104) operatively coupled to the camera (102) and adapted to receive the captured live video; an input conditioning unit (106) adapted to process the received video by performing frame generation and quality enhancement; a processing unit (108) operatively coupled to the input conditioning unit (106), characterized in that the processing unit (108) is configured to: receive the live video stream of the face of the user during device usage; process the received video stream by performing frame generation and quality enhancement; detect a facial region and an eye region using a deep learning-based face detection model in each frame; extract blink-related features from the detected eye region, wherein the blink-related features comprise a blink frequency, a blink duration, and blink irregularities or a combination thereof; generate a feature vector based on the extracted blink-related features over a predefined time interval; input the feature vector into a trained Multilayer Perceptron (MLP) classifier; classify the addiction risk level of the user into predefined categories comprising normal, moderate, or high risk via the multilayer perceptron classifier; and display the classified addiction risk level along with corresponding wellness feedback on the user device.

2. The system (100) as claimed in claim 1, comprising a decision unit (110) adapted to determine addiction risk level based on the prediction and an output interface (112) adapted to present the determined addiction risk level to the user.

3. The system (100) as claimed in claim 1, wherein the processing unit (108) is configured to execute a Retina-Face-based model to detect facial landmarks for precise eye localization.

4. The system (100) as claimed in claim 1, wherein the eye region of interest is dynamically tracked across successive frames.

5. The system (100) as claimed in claim 1, wherein the blink frequency is computed as a number of blinks per predefined time interval.

6. The system (100) as claimed in claim 1, wherein the blink duration corresponds to a time interval of eyelid closure.

7. The system (100) as claimed in claim 1, comprising a feedback unit (114) configured to generate personalized wellness suggestions based on addiction risk level.

8. A computer-implemented method (500) for predicting an electronic device addiction, the method (500) is characterized by steps of: capturing a live video stream of a face of the user during device usage through a camera (102) of the electronic device; processing the received video stream by performing frame generation and quality enhancement; detecting a facial region and an eye region using a deep learning-based face detection model in each frame; extracting blink-related features from a detected eye region, wherein the blink-related features comprise a blink frequency, a blink duration, and blink irregularities or a combination thereof; generating a feature vector based on the extracted blink-related features over a predefined time interval; and inputting the feature vector into a trained Multilayer Perceptron (MLP) classifier.

9. The method (500) as claimed in claim 8, comprising a step of classifying the addiction risk level of the user into predefined categories comprising normal, moderate, or high risk via the multilayer perceptron classifier.

10. The method (500) as claimed in claim 8, comprising a step of displaying the classified addiction risk level along with corresponding wellness feedback on the user device. Date: March 09, 2026 Place: Noida Nainsi Rastogi Patent Agent (IN/PA-2372) Agent for the Applicant

Description:BACKGROUND
Field of Invention
[001] Embodiments of the present invention generally relate to digital wellness and user behaviour assessment systems and particularly to a system and method for predicting an electronic device addiction of a user.
Description of Related Art
[002] Widespread electronic device use creates health and behavioural concerns. Many users show eye strain, fatigue, reduced attention span, and compulsive device use. Clinicians and families face difficulty in early identification of harmful digital habits because clear physiological indicators and objective measures remain limited. Reliance on self-reports and simple device statistics does not reflect actual cognitive or visual stress.
[003] Several digital wellness tools exist in market. Screen time applications record total duration, number of unlocks, and app access history. Platform dashboards from major mobile platforms allow timers, alerts, and usage limits. Some research systems rely on dedicated eye tracker hardware or clinical assessment methods.
[004] These solutions present several limitations. Screen time data alone fails to reveal physiological state or mental fatigue. Manual self-reports suffer from bias and low reliability. Specialized eye tracker devices require high cost, complex setup, and controlled environments. As a result, accurate early detection of problematic electronic device use remains difficult.
[005] There is thus a need for an improved and advanced system and method for predicting an electronic device addiction of a user that can administer the aforementioned limitations in a more efficient manner.
SUMMARY
[006] Embodiments in accordance with the present invention provide a system for predicting an electronic device addiction of a user. The system comprising a camera, adapted to capture a live video stream of a face of a user. The camera is built into the electronic device. The system further comprising an input interface operatively coupled to the camera and adapted to receive the captured live video. The system further comprising an input conditioning unit adapted to process the received video by performing frame generation and quality enhancement. The system further comprising a processing unit operatively coupled to the input conditioning unit. The processing unit is configured to receive the live video stream of the face of the user during device usage; process the received video stream by performing frame generation and quality enhancement; detect a facial region and an eye region using a deep learning-based face detection model in each frame; extract blink-related features from the detected eye region. The blink-related features comprise a blink frequency, a blink duration, and blink irregularities or a combination thereof; generate a feature vector based on the extracted blink-related features over a predefined time interval; input the feature vector into a trained Multilayer Perceptron (MLP) classifier; classify the addiction risk level of the user into predefined categories comprising normal, moderate, or high risk via the multilayer perceptron classifier; and display the classified addiction risk level along with corresponding wellness feedback on the user device.
[007] Embodiments in accordance with the present invention further provide a method for predicting an electronic device addiction. The method comprising steps of capturing a live video stream of a face of the user during device usage through a camera of the electronic device; processing the received video stream by performing frame generation and quality enhancement; detecting a facial region and an eye region using a deep learning-based face detection model in each frame; extracting blink-related features from a detected eye region. The blink-related features comprise a blink frequency, a blink duration, and blink irregularities or a combination thereof; generating a feature vector based on the extracted blink-related features over a predefined time interval; and inputting the feature vector into a trained Multilayer Perceptron (MLP) classifier.
[008] Embodiments of the present invention may provide a number of advantages depending on their particular configuration. First, embodiments of the present application may provide a system for predicting an electronic device addiction.
[009] Next, embodiments of the present application may provide a system that enables early identification of an electronic device addiction risk through physiological eye behaviour analysis rather than reliance on self-reported usage data.
[0010] Next, embodiments of the present application may provide a system that provides real time addiction assessment using camera-based monitoring without requiring specialized eye-tracking hardware.
[0011] Next, embodiments of the present application may provide a system that improves prediction accuracy through deep-learning-based analysis of blink frequency, duration, and irregularity.
[0012] Next, embodiments of the present application may provide a system that supports personalized digital wellness intervention by generating user-specific feedback based on detected addiction risk levels.
BRIEF DESCRIPTION OF THE DRAWINGS
[0013] FIG. 1 illustrates a block diagram of a system for predicting an electronic device addiction of a user, according to an embodiment of the present invention;
[0014] FIG. 2 illustrates a block diagram of a processing unit of the system for predicting the electronic device addiction of the user, according to an embodiment of the present invention;
[0015] FIG. 3A illustrates a workflow diagram representing an addiction prediction pipeline of the system for predicting the electronic device addiction of the user, according to an embodiment of the present invention;
[0016] FIG. 3B illustrates a detailed architectural flow diagram of the system for predicting the electronic device addiction of the user, according to an embodiment of the present invention;
[0017] FIG. 4A illustrates a home interface of an output interface, according to an embodiment of the present invention;
[0018] FIG. 4B illustrates an operational home interface in an active state diagram of the output interface, according to an embodiment of the present invention;
[0019] FIG. 4C illustrates a real time dashboard interface of the output interface, according to an embodiment of the present invention;
[0020] FIG. 4D illustrates a real time prediction display interface of the output interface, according to an embodiment of the present invention;
[0021] FIG. 4E illustrates a real time evaluation interface of the output interface, according to an embodiment of the present invention;
[0022] FIG. 4F illustrates a normal addiction evaluation interface of the output interface, according to an embodiment of the present invention;
[0023] FIG. 4G illustrates a blink timeline interface of the output interface, according to an embodiment of the present invention;
[0024] FIG. 4H illustrates a settings interface of the output interface, according to an embodiment of the present invention; and
[0025] FIG. 5 depicts a flowchart of a method for predicting the electronic device addiction, according to an embodiment of the present invention.
DETAILED DESCRIPTION
[0026] As used herein, the term “user” may refer to an individual, person, operator, device holder, account holder, or any human interacting with an electronic device during an operation. The user may be, but not limited to, a primary device owner, a secondary device operator, a minor, an adult, a patient, a student, a professional, or any person engaging with applications, interfaces, multimedia content, communication platforms, or other digital environments, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the user interacting with the electronic device, including known usage contexts, shared-device environments, supervised-use scenarios, and/or later developed interaction paradigms.
[0027] FIG. 1 illustrates a block diagram of a system 100 for predicting an electronic device addiction of a user, according to an embodiment of the present invention. In an embodiment of the present invention, the system 100 may be deployed on the electronic device and may be adapted to monitor usage behaviour of the electronic device during real time interaction of the user with applications. The electronic device may be, but not limited to, a smartphone, a laptop, a desktop computer, a gaming console, and so forth.
[0028] The system 100 may be adapted to deploy/activate a front-facing camera adapted to capture facial and ocular behaviour while the user engages with applications, browsing interfaces, multimedia content, or communication platforms. The system 100 may be configured to operate in a background mode during usage sessions and may continuously analyse blink-related physiological parameters indicative of smartphone addiction patterns.
[0029] In an embodiment of the present invention, the system 100 may provide automated and objective assessment of the electronic device addiction based on physiological eye behaviour analysis. The system 100 may enable real time monitoring and predictive evaluation using computer vision and machine learning techniques to improve reliability of digital wellness assessment. The system 100 may operate in a continuous or on-demand manner during user interaction with the electronic device. The system 100 may support scalable deployment across consumer electronic devices and may provide adaptive and reliable addiction risk assessment under natural usage conditions.
[0030] In an embodiment of the present invention, the system 100 may be adapted for low computational overhead and scalable deployment across mass-market consumer electronic devices. The system 100 may utilize optimized deep-learning architectures and may be configured to enable real time processing without requiring dedicated high-performance graphics processing hardware. The system 100 may therefore be suitable for portable, low-cost, and large-scale deployment for electronic device addiction assessment.
[0031] According to the embodiments of the present invention, the system 100 may incorporate non-limiting hardware components to enhance the processing speed and efficiency such as the system 100 may comprise a camera 102, an input interface 104, an input conditioning unit 106, a processing unit 108, a decision unit 110, an output interface 112, a feedback unit 114, and an alert unit 116. In an embodiment of the present invention, the hardware components of the system 100 may be integrated with computer-executable instructions for overcoming the challenges and the limitations of the existing systems.
[0032] In an embodiment of the present invention, the camera 102 may be adapted to capture live video of the user through the camera 102 of the electronic device. The camera 102 may include a built-in front-facing camera or any imaging sensor capable of facial video capture. The captured live video may include facial and eye-region information indicative of user eye behaviour. Such information may reflect physiological responses associated with fatigue, visual strain, or prolonged device engagement.
[0033] In an embodiment of the present invention, the system 100 may be adapted to operate using the camera 102 inherently available within the electronic device without requiring additional external eye-tracking hardware, infrared sensors, wearable tracking devices, structured-light systems, or specialized ophthalmic imaging equipment. The camera 102 may comprise a built-in front-facing camera of a smartphone, tablet, laptop, desktop computer, or any consumer electronic device capable of capturing video data. In an embodiment of the present invention, the processing unit 108 may be adapted to perform eye detection and blink-related feature extraction under natural ambient lighting conditions without dependency on artificial illumination modules. Embodiments of the present invention may include operation on consumer-grade imaging devices without hardware modification, including known, related art, and/or later developed camera technologies.
[0034] The camera 102 may be, but not limited to, a built-in front-facing camera, a rear-facing camera, a complementary metal-oxide semiconductor imaging sensor, a charge-coupled device imaging sensor, a depth camera, or any imaging module capable of capturing facial video data, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the camera 102, including known, related art, and/or later developed imaging technologies.
[0035] In an embodiment of the present invention, the input interface 104 may be operatively coupled to the camera 102 and adapted to receive the captured live video. The input interface 104 may facilitate transmission of video data to downstream processing components.
[0036] The input interface 104 may be, but not limited to, a video acquisition interface, a camera driver interface, an application programming interface, a hardware abstraction interface, a data buffering interface, or any communication interface configured to receive video data from the camera 102, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the input interface 104, including known, related art, and/or later developed interface technologies.
[0037] The received video may be transmitted to the input conditioning unit 106. The input conditioning unit 106 may be adapted to process the received video by performing frame generation and quality enhancement to generate conditioned frame data. The processing may be, but not limited to, a frame extraction, a normalization, an illumination correction, a noise reduction, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of preprocessing techniques, including known, related art, and/or later developed technologies.
[0038] The input conditioning unit 106 may be, but not limited to, a preprocessing engine, a frame extraction module, an image normalization module, a noise reduction module, a filtering module, or any video conditioning mechanism adapted to enhance captured frame quality, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the input conditioning unit 106, including known, related art, and/or later developed preprocessing technologies.
[0039] The conditioned frame data may be transmitted to the processing unit 108 operatively coupled to the input conditioning unit 106 and a memory storing executable instructions. The processing unit 108 may be configured to capture a live video stream of a face of the user during device usage through the camera 102 using a Retina-Face-based model. The Retina-Face-based model may enable accurate facial landmark detection and eye localization under varying lighting and head-position conditions. The processing unit 108 may process the received video stream by performing frame generation and quality enhancement and detect a facial region and an eye region using a deep learning-based face detection model in each frame.
[0040] The processing unit 108 may further be configured to extract blink-related features from a detected eye region. The blink-related features may comprise a blink frequency, a blink duration, and blink irregularities, and so forth. The processing unit 108 may be configured to generate a feature vector based on the extracted blink-related features over a predefined time interval. The blink frequency may represent a number of blinks within a predefined time interval. The blink duration may represent a time interval between eye closure and reopening. The blink irregularities may represent variation in blink intervals across time. Embodiments of the present invention are intended to include or otherwise cover any variations of blink-feature computation.
[0041] In an embodiment of the present invention, the processing unit 108 may be adapted to perform continuous behavioural modelling by analysing the blink-related features across successive frames over dynamic time windows. The processing unit 108 may be configured to update a generated feature vector at predefined or adaptive intervals to capture evolving blink behaviour patterns. The addiction classification module 206 may be adapted to evaluate temporal variations, deviation trends, progressive irregularity, and frequency shifts in blink behaviour to identify early-stage addiction indicators. Embodiments of the present invention may include continuous real time adaptive behavioural modelling mechanisms.
[0042] The processing unit 108 may input the feature vector into a trained Multilayer Perceptron (MLP) classifier. The Multilayer Perceptron (MLP) classifier may analyse blink-feature patterns to generate an addiction-level prediction. The addiction-level prediction may correspond to predefined categories such as normal, moderate, or high addiction risk. Embodiments of the present invention are intended to include or otherwise cover any machine learning models suitable for behavioural classification.
[0043] In an embodiment of the present invention, the processing unit 108 may be adapted to perform addiction-level prediction based primarily on physiological blink-related parameters rather than relying solely on device usage logs, screen time duration metrics, application access history, or self-reported behavioural data. The processing unit 108 may thereby provide an objective biometric assessment of electronic device addiction using eye behaviour analysis derived from captured video frames. Embodiments of the present invention may include biometric-driven behavioural evaluation systems.
[0044] The processing unit 108 may be, but not limited to, a central processing unit, a graphics processing unit, a neural processing unit, a digital signal processor, a system-on-chip architecture, a cloud-based computing engine, or any computational hardware or virtual processing platform configured to execute machine-learning operations, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the processing unit 108, including known, related art, and/or later developed processing technologies. The processing unit 108 may further be explained in detail in conjunction with FIG. 2.
[0045] The addiction-level prediction may be transmitted to the decision unit 110. The decision unit 110 may be adapted to determine a final addiction risk level based on prediction outputs and predefined thresholds. The decision unit 110 may support confidence evaluation and categorization logic. The decision unit 110 may be, but not limited to, a rule-based evaluation engine, a threshold comparator module, a probabilistic scoring engine, a classification refinement module, or any logical decision-making framework adapted to determine addiction risk levels, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the decision unit 110, including known, related art, and/or later developed decision-processing technologies.
[0046] The determined addiction risk level may be transmitted to the output interface 112. The output interface 112 may be adapted to present addiction status for visualization, monitoring, or notification. The output interface 112 may support dashboards, mobile application displays, or alert generation. Embodiments of the present invention are intended to include or otherwise cover any type of output formats or presentation mechanisms.
[0047] The output interface 112 may be, but not limited to, a graphical user interface, a mobile application interface, a web-based interface, a dashboard display module, a notification panel, or any visual or interactive presentation interface configured to present addiction assessment results, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the output interface 112, including known, related art, and/or later developed output presentation technologies. The output interface 112 may further be explained in detail in conjunction with FIG. 4A to FIG. 4H.
[0048] In an embodiment of the present invention, the system 100 may further comprise the feedback unit 114 adapted to generate personalized wellness recommendations based on addiction risk level. The recommendations may include break reminders, screen-use reduction suggestions, or eye-rest notifications. Embodiments of the present invention are intended to include or otherwise cover any variations of feedback generation mechanisms. The feedback unit 114 may be, but not limited to, a recommendation engine, a behavioural guidance module, a digital wellness advisory system, a personalized suggestion generator, or any intervention-generation mechanism configured to provide corrective guidance based on addiction risk levels, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the feedback unit 114, including known, related art, and/or later developed behavioural feedback technologies.
[0049] In an embodiment of the present invention, the alert unit 116 may be operatively coupled to the decision unit 110 and may be adapted to receive the determined addiction risk level from the decision unit 110. The alert unit 116 may be configured to generate alert signals corresponding to the classified addiction risk level. The alert signals may include visual notifications, auditory alerts, vibration signals, pop-up warnings, banner messages, or combinations thereof displayed on the electronic device. In an embodiment of the present invention, the alert unit 116 may be adapted to trigger graduated alert responses based on predefined risk thresholds, persistence duration, or recurrence frequency of high-risk classifications. The alert unit 116 may further be configured to interface with the output interface 112 to present contextual intervention prompts that may be, but are not limited to, break reminders, usage limitation suggestions, temporary screen-lock mechanisms, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of adaptive alert generation mechanisms suitable for behavioural intervention and addiction mitigation.
[0050] The alert unit 116 may be, but not limited to, a visual notification generator, an auditory alert generator, a vibration trigger module, a pop-up warning module, a banner notification system, or any alert generation mechanism adapted to provide real time addiction risk notifications, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the alert unit 116, including known, related art, and/or later developed alerting technologies.
[0051] FIG. 2 illustrates components of the processing unit 108 of the system 100 according to an embodiment of the present invention. The processing unit 108 may comprise a data processing module 200, a detection module 202, a feature extraction module 204, an addiction classification module 206, and a feedback module 208.
[0052] In an embodiment of the present invention, the data processing module 200 may be configured to receive the live video stream captured through the camera 102. The data processing module 200 may be configured to convert the received video stream into frames. The data processing module 200 may be further configured to perform frame enhancement such as, but not limited to, a brightness normalization, a contrast adjustment, a noise reduction, a resolution refinement, and so forth. In an embodiment of the present invention, the data processing module 200 may be configured to filter low-quality frames based on predefined quality parameters. Further, the data processing module 200 may be configured to transmit processed frames to the detection module 202.
[0053] In an embodiment of the present invention, the detection module 202 may be configured to detect a facial region and isolate an eye region in each frame using a deep learning-based face detection model. The deep learning-based face detection model may be, but not limited to, Retina-Face-based model, or equivalent neural network architectures. In an embodiment of the present invention, the detection module 202 may be configured to identify facial landmarks corresponding to eye boundaries. Further, the detection module 202 may be configured to transmit the isolated eye region to the feature extraction module 204.
[0054] The feature extraction module 204 may be configured to receive the isolated eye region from the detection module 202 and determine blink-related physiological parameters. The blink-related parameters may comprise the blink frequency, the blink duration, the blink irregularities, and so forth. The feature extraction module 204 may be configured to compute the blink frequency based on a number of blink events detected within a predefined time interval. Further, the feature extraction module 204 may be configured to determine the blink duration based on time intervals corresponding to eye closure and reopening. The feature extraction module 204 may also be configured to determine the blink irregularities based on variation between consecutive blink intervals. In an embodiment of the present invention, the feature extraction module 204 may be configured to generate a structured feature vector and transmit the generated feature vector to the addiction classification module 206.
[0055] In an embodiment of the present invention, the addiction classification module 206 may be configured to receive the feature vector from the feature extraction module 204. The addiction classification module 206 may be configured to input the feature vector into a trained Multilayer Perceptron (MLP) classifier. The Multilayer Perceptron (MLP) classifier may comprise hidden layers configured to identify nonlinear relationships between the blink-related features and electronic device addiction patterns. In an exemplary scenario, if the addiction classification module 206 determines that the evaluated feature vector corresponds to predefined low-risk parameters, then the addiction classification module 206 may be configured to classify the addiction risk level as normal. In another exemplary scenario, if the evaluated feature vector corresponds to moderate deviation parameters, then the addiction classification module 206 may be configured to classify the addiction risk level as moderate. In yet another exemplary scenario, if the evaluated feature vector corresponds to high-risk physiological patterns, then the addiction classification module 206 may be configured to classify the addiction risk level as high risk. Further, the addiction classification module 206 may be configured to transmit the classified addiction risk level to the feedback module 208.
[0056] The feedback module 208 may be configured to receive the classified addiction risk level from the addiction classification module 206. The feedback module 208 may be configured to generate user notifications corresponding to the classified addiction risk level. In an embodiment of the present invention, if the classified addiction risk level is normal, then the feedback module 208 may be configured to display a status notification indicating healthy usage. In another embodiment of the present invention, if the classified addiction risk level is moderate, then the feedback module 208 may be configured to generate a recommendation suggesting reduction of continuous screen usage.
[0057] In an embodiment of the present invention, if the classified addiction risk level is high risk, then the feedback module 208 may be configured to generate an alert suggesting immediate break or screen time control measures. The feedback module 208 may be configured to display the classified addiction risk level along with personalized wellness guidance on the user interface of the device.
[0058] In an embodiment of the present invention, the feedback module 208 and the alert unit 116 may be adapted to generate preventive health advisories associated with prolonged electronic device usage. The advisories may correspond to physiological indicators such as eye strain, fatigue, cognitive overload, dry-eye conditions, headache risk, reduced attention span, and stress-related symptoms inferred from blink-related parameters. The system 100 may be configured to provide early-warning notifications encouraging behavioural correction prior to manifestation of long-term health complications. Embodiments of the present invention may include preventive digital health management and behavioural intervention mechanisms.
[0059] FIG. 3A illustrates a workflow diagram 300 of the system 100 representing an addiction prediction pipeline, according to an embodiment of the present invention. The workflow diagram 300 may depict exemplary sequential processing stages for analysing user eye behaviour and generating addiction alerts. The workflow diagram 300 may begin with acquisition of a video frame obtained from the camera 102. The video frame may include facial and eye-region information of a user during device interaction. The video frames may serve as primary input for further analysis. The video frame data may be transmitted to a feature extraction stage. The feature extraction stage may derive relevant eye behaviour attributes from captured frames. Such attributes may correspond to eye movement characteristics and blink-related parameters. The extracted features may be forwarded to a blink analysis stage. The blink analysis stage may evaluate the blink frequency, the blink duration, and the blink irregularities across successive frames. The blink analysis may identify behavioural indicators associated with fatigue or excessive device use.
[0060] Outputs from blink analysis may be transmitted to a classification stage. The classification stage may utilize machine-learning models to categorize user behaviour into predefined addiction-risk levels. The classification results may be transmitted to the alert unit 116. The alert unit 116 may generate notifications or alerts based on the predicted addiction risk level. The alerts may appear on a user device and may prompt behavioural intervention. Embodiments of the present invention are intended to include or otherwise cover any type of alert mechanisms, including known, related art, and/or later developed technologies. In an embodiment of the present invention, the alert unit 116 may be adapted to receive classification results from the prediction module and determine an appropriate alert response. The alert unit 116 may analyse the predicted addiction risk level in conjunction with predefined rules, thresholds, user profiles, or contextual information to generate customized alerts. Such alerts may include visual, auditory, haptic, or textual notifications, and may further be adapted based on the severity, frequency, or duration of the detected addictive behaviour. The alert unit 116 may also be configured to escalate alerts, trigger follow-up actions, or interface with external systems, caregivers, or intervention platforms to facilitate timely behavioural intervention.
[0061] In an embodiment of the present invention, the alert unit 116 may be adapted to escalate notifications based on severity, persistence, frequency, or recurrence of high-risk addiction classifications. The alert unit 116 may be configured to generate repeated alerts, extended break recommendations, adaptive intervention prompts, or optional transmission of summarized addiction assessment reports to authorized supervisory systems or caregiver interfaces, subject to user-configurable privacy settings. Embodiments of the present invention may include multi-level behavioural intervention frameworks based on addiction risk progression.
[0062] FIG. 3B illustrates a detailed architectural flow diagram 302 of the system 100, according to an embodiment of the present invention. The architecture flow diagram 302 may depict layered processing of video input for addiction prediction. The process may begin within an image-processing stage employing a Retina-Face-based engine 304. A video frame from the feed of the camera 102 may be supplied to the Retina-Face-based engine 304. The Retina-Face-based engine 304 may detect facial landmarks and eye landmarks for precise eye localization. The localized eye regions may be transmitted to a cloud or intermediate processing stage for further analysis. The cloud stage may facilitate data transfer, storage, or distributed processing.
[0063] The process may further employ an AI-based decision engine 306. The AI-based decision engine 306 may receive localized eye crops derived from detected eye regions. In an embodiment of the present invention, the eye crops may be processed using neural-network-based models such as, but not limited to, MobileNetV2 or other vision models suitable for eye-region analysis. Embodiments of the present invention are intended to include or otherwise cover any type of neural-network architectures. The AI-based decision engine 306 may generate a blink timeline over a predefined duration, such as a two-minute window. The blink timeline may represent temporal blink behaviour of the user.
[0064] Feature extraction may then derive fatigue-related or addiction-related indicators from the blink timeline. The derived features may support behavioural analysis and prediction. The processed outputs may be transmitted to the alert unit 116, that may categorize addiction risk as low, medium, or high. The alert unit 116 may present notifications or visual indicators to the user. Embodiments of the present invention are intended to include or otherwise cover any type of alert presentation mechanisms.
[0065] FIG. 4A illustrates a home interface 400 of the output interface 112, according to an embodiment of the present invention. The home interface 400 may be rendered via a web application framework and may provide an overview of addiction detection functionality based on blink physiology. The home interface 400 may display introductory information describing the blink frequency, the blink duration, and the blink irregularities as indicators for addiction assessment. The home interface 400 may include a navigation bar positioned on an upper portion of the display, providing selectable options such as, but not limited to, Home, Dashboard, Settings, and so forth. The home interface may further include selectable controls such as “Start Application” and “Open Dashboard” to initiate monitoring or access analytics. The home interface may display sample or current values related to the blink frequency, the blink duration, and the blink irregularities. Embodiments of the present invention are intended to include or otherwise cover any type of home-screen visualization or navigation arrangement.
[0066] FIG. 4B illustrates an operational home interface 402 in an active state of the output interface 112, according to an embodiment of the present invention. The operational home interface 402 in an active state may present real time blink measurements during operation of the system 100. The operational home interface 402 in an active state may display a running status indicator along with dynamically updated blink metrics such as, but not limited to, blinks per minute, the blink duration, and irregularity values. The interface may further include controls for stopping the application or navigating to analytical dashboards. In an embodiment of the present invention, the operational home interface 402 in active state may support automatic refresh of blink-related parameters at predefined intervals. The operational home interface 402 in active state may enable visualization of temporal changes in blink behaviour to assist in monitoring user eye activity over time. The operational home interface 402 in active state may further support notification indicators when measured values exceed predefined thresholds. Embodiments of the present invention are intended to include or otherwise cover any type of real time interface updates or notification mechanisms.
[0067] FIG. 4C illustrates a real time dashboard interface 404 of the output interface 112, according to an embodiment of the present invention. The real time dashboard interface 404 may provide the live video preview region for capturing user facial data. The dashboard may include selectable controls such as “Start Camera” and “Stop” to manage video acquisition. The dashboard may further display model status indicators and sections for prediction results and feature summaries. The prediction section may display addiction risk level outputs and confidence indicators. The feature summary section may display blink-related statistics such as, but not limited to, the blink frequency, average blink duration, irregularity measures, and so forth. In an embodiment of the present invention, the real time dashboard interface 404 may enable user interaction for initiating or terminating data capture sessions. The real time dashboard interface 404 may further support historical data visualization and session-based monitoring. The live preview region may assist in proper user positioning for accurate eye detection. Embodiments of the present invention are intended to include or otherwise cover any type of user-interactive dashboard functionality.
[0068] FIG. 4D illustrates a real time prediction display interface 406 of the output interface 112, according to an embodiment of the present invention. The real time prediction display interface 406 may display prediction results and present an addiction-level classification such as high addiction risk along with an associated confidence value. The real time prediction display interface 406 may further display blink-related feature summaries such as, but not limited to, blinks per minute, average duration, and irregularity values. A blink timeline section may provide temporal visualization of blink events. In an embodiment of the present invention, the real time prediction display interface 406 may enable storage or logging of prediction outcomes for longitudinal assessment. The blink timeline visualization may assist in identifying behavioural trends associated with prolonged device usage. The real time prediction display interface 406 may further support visual indicators or color-coded alerts for intuitive interpretation of addiction risk levels. Embodiments of the present invention are intended to include or otherwise cover any type of prediction visualization or logging features.
[0069] FIG. 4E illustrates a real time evaluation interface 408 of the output interface 112, according to an embodiment of the present invention. The real time evaluation interface 408 may present a moderate addiction evaluation result. The real time evaluation interface 408 may include a video preview region that presents a camera initialization state. Control elements such as a “Start Camera” control and a “Stop” control may support user control of video acquisition. A model status indicator may present a running state of the system 100. A prediction panel may present an addiction risk level classification of moderate along with a confidence value. A feature summary panel may present blink metrics such as, but not limited to, a blink frequency per minute, an average blink duration, irregularity index, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of dashboard interface that presents real time physiological indicators and classification outputs.
[0070] FIG. 4F illustrates a normal addiction evaluation interface 410 of the output interface 112, according to an embodiment of the present invention. The normal addiction evaluation interface 410 may include a live preview region associated with the feed of the camera 102. The normal addiction evaluation interface 410 may further include operational controls for activation and termination of the camera 102. A prediction panel may present a normal addiction risk level along with a corresponding confidence score. A feature summary panel may present the blink frequency, the blink duration, and the blink irregularities values. The normal addiction evaluation interface 410 may provide a structured layout that enables a user to interpret physiological parameters and classification outputs. Embodiments of the present invention are intended to include or otherwise cover any type of dashboard that presents categorized addiction assessment results.
[0071] FIG. 4G illustrates a blink timeline interface 412 of the output interface 112, according to an embodiment of the present invention. The blink timeline interface 412 may present chronological blink event records. The blink timeline interface 412 may list time-stamped blink entries along with the blink frequency, the blink duration, and irregularity values. The timeline may support review of blink behaviour across successive time intervals. The blink timeline interface 412 may enable visualization of temporal blink patterns that support addiction risk level evaluation. Embodiments of the present invention are intended to include or otherwise cover any type of chronological visualization interface for physiological event tracking.
[0072] FIG. 4H illustrates a settings interface 414 of the output interface 112, according to an embodiment of the present invention. The settings interface 414 may enable parameter configuration. The settings interface 414 may include editable threshold fields such as a high blink frequency threshold, a moderate blink frequency threshold, and a time window parameter. A save control may enable storage of customized settings. The settings interface 414 may further display reference sample values for blink metrics and predicted addiction risk level. The settings interface 414 may support user-defined calibration of evaluation criteria. Embodiments of the present invention are intended to include or otherwise cover any type of configurable parameter interface for physiological assessment systems. In an embodiment of the present invention, the settings interface 414 may be adapted to allow user-defined configuration of blink-frequency thresholds, blink-duration thresholds, irregularity-index thresholds, and time-window parameters used by the processing unit 108 and decision unit 110 for addiction-level classification. The system 100 may be configured to perform adaptive calibration based on user-specific baseline blink behaviour collected over an initial observation period. Embodiments of the present invention may include customizable and adaptive physiological evaluation criteria.
[0073] FIG. 5 illustrates a flowchart of a method 500 for predicting the electronic device addiction using the system 100 according to an embodiment of the present invention. At step 502, the system 100 may capture the live video stream of the face of the user during device usage through the camera 102.
[0074] At step 504, the system 100 may process the received video stream by performing the frame generation and the quality enhancement. At step 506, the system 100 may detect the face and the eye region using the face detection model in each frame.
[0075] At step 508, the system 100 may extract the blink-related features from a detected eye region. The blink-related features may comprise the blink frequency, the blink duration, the blink irregularities, and so forth.
[0076] At step 510, the system 100 may generate the feature vector based on the extracted blink-related features over the predefined time interval. At step 512, the system 100 may input the feature vector into the trained Multilayer Perceptron (MLP) classifier.
[0077] At step 514, the system 100 may classify the addiction risk level of the user into predefined categories comprising normal, moderate, or high risk via the multilayer perceptron classifier.
[0078] At step 516, the system 100 may display the classified addiction risk level along with corresponding wellness feedback on the user device. , Claims:CLAIMS
I/We Claim:
1. A system (100) for predicting an electronic device addiction of a user, the system (100) comprising:
a camera (102), adapted to capture a live video stream of a face of a user, wherein the camera (102) is built into the electronic device;
an input interface (104) operatively coupled to the camera (102) and adapted to receive the captured live video;
an input conditioning unit (106) adapted to process the received video by performing frame generation and quality enhancement;
a processing unit (108) operatively coupled to the input conditioning unit (106), characterized in that the processing unit (108) is configured to:
receive the live video stream of the face of the user during device usage;
process the received video stream by performing frame generation and quality enhancement;
detect a facial region and an eye region using a deep learning-based face detection model in each frame;
extract blink-related features from the detected eye region, wherein the blink-related features comprise a blink frequency, a blink duration, and blink irregularities or a combination thereof;
generate a feature vector based on the extracted blink-related features over a predefined time interval;
input the feature vector into a trained Multilayer Perceptron (MLP) classifier;
classify the addiction risk level of the user into predefined categories comprising normal, moderate, or high risk via the multilayer perceptron classifier; and
display the classified addiction risk level along with corresponding wellness feedback on the user device.
2. The system (100) as claimed in claim 1, comprising a decision unit (110) adapted to determine addiction risk level based on the prediction and an output interface (112) adapted to present the determined addiction risk level to the user.
3. The system (100) as claimed in claim 1, wherein the processing unit (108) is configured to execute a Retina-Face-based model to detect facial landmarks for precise eye localization.
4. The system (100) as claimed in claim 1, wherein the eye region of interest is dynamically tracked across successive frames.
5. The system (100) as claimed in claim 1, wherein the blink frequency is computed as a number of blinks per predefined time interval.
6. The system (100) as claimed in claim 1, wherein the blink duration corresponds to a time interval of eyelid closure.
7. The system (100) as claimed in claim 1, comprising a feedback unit (114) configured to generate personalized wellness suggestions based on addiction risk level.
8. A computer-implemented method (500) for predicting an electronic device addiction, the method (500) is characterized by steps of:
capturing a live video stream of a face of the user during device usage through a camera (102) of the electronic device;
processing the received video stream by performing frame generation and quality enhancement;
detecting a facial region and an eye region using a deep learning-based face detection model in each frame;
extracting blink-related features from a detected eye region, wherein the blink-related features comprise a blink frequency, a blink duration, and blink irregularities or a combination thereof;
generating a feature vector based on the extracted blink-related features over a predefined time interval; and
inputting the feature vector into a trained Multilayer Perceptron (MLP) classifier.
9. The method (500) as claimed in claim 8, comprising a step of classifying the addiction risk level of the user into predefined categories comprising normal, moderate, or high risk via the multilayer perceptron classifier.
10. The method (500) as claimed in claim 8, comprising a step of displaying the classified addiction risk level along with corresponding wellness feedback on the user device.
Date: March 09, 2026
Place: Noida

Nainsi Rastogi
Patent Agent (IN/PA-2372)
Agent for the Applicant