Description:FIELD OF DISCLOSURE
[0001] The present disclosure relates generally relates artificial intelligence and machine learning applications in healthcare diagnostics, more specifically, relates to brain stroke prediction system and method thereof.
BACKGROUND OF THE DISCLOSURE
[0002] The present disclosure is integrating explainable artificial intelligence techniques, such as shapely additive explanations and local interpretable model-agnostic explanations, into stroke prediction models. This integration is providing healthcare professionals with transparent insights into the model's decision-making process. By elucidating the contribution of each feature to the prediction, clinicians understand the underlying factors influencing stroke risk assessments. This transparency is fostering trust in the system's outputs and is facilitating informed clinical decisions, thereby enhancing patient care.
[0003] The present disclosure is employing Internet of Things devices, including wearable sensors, to continuously monitor patients' vital signs such as heart rate variability, blood pressure, and oxygen saturation. These devices are collecting real-time data, which is being a processed locally using edge computing technique. This approach is enabling immediate analysis and detection of abnormal patterns indicative of stroke risk. The system is promptly alerting healthcare providers, allowing for swift intervention and potentially mitigating the severity of stroke events.
[0004] The present disclosure is utilizing federated learning methodologies to train predictive models across multiple healthcare institutions without sharing raw patient data. Each institution is training the model locally on its data and is sharing only the model updates. This decentralized approach is preserving patient privacy and is complying with data protection regulations. Simultaneously, it is enabling the aggregation of diverse datasets, enhancing the model's generalizability and accuracy across different populations and healthcare settings.
[0005] Existing stroke prediction systems are often relying on complex machine learning algorithms that function as "black boxes," providing little to no insight into how predictions are made. This lack of interpretability is hindering clinicians' ability to understand and trust the model's outputs. Consequently, healthcare providers are facing challenges in validating the system's recommendations, which is potentially impacting the adoption and effectiveness of such technologies in clinical practice.
[0006] Many current models are trained on datasets where non-stroke cases significantly outnumber stroke cases, leading to class imbalance. These models are often failing to accurately identify high-risk patients, resulting in a higher rate of false negatives. The inability to effectively handle imbalanced data is compromising the reliability of stroke predictions and is posing a risk to patient safety by potentially overlooking individuals who are at genuine risk of experiencing a stroke.
[0007] Traditional stroke prediction systems are requiring centralized data collection, necessitating the transfer of sensitive patient information to a central repository. This approach is raising significant privacy concerns and is often conflicting with data protection laws and regulations. Additionally, integrating data from disparate sources is presenting technical challenges, including issues with data standardization and interoperability, which are impeding the development of comprehensive and accurate predictive models.
[0008] Thus, in light of the above-stated discussion, there exists a need for a brain stroke prediction system and method thereof.
SUMMARY OF THE DISCLOSURE
[0009] The following is a summary description of illustrative embodiments of the invention. It is provided as a preface to assist those skilled in the art to more rapidly assimilate the detailed design discussion which ensues and is not intended in any way to limit the scope of the claims which are appended hereto in order to particularly point out the invention.
[0010] According to illustrative embodiments, the present disclosure focuses on brain stroke prediction system and method thereof which overcomes the above-mentioned disadvantages or provides the users with a useful or commercial choice.
[0011] An objective of the present disclosure is to develop an artificial intelligence-based system that is continuously analysing patient-specific data to detect early indicators of brain stroke risk with enhanced precision, thereby enabling timely medical interventions and reducing neurological damage.
[0012] An objective of the present disclosure is to integrate explainable artificial intelligence methodologies into the stroke prediction system for generating transparent, interpretable, and clinically actionable insights that are assisting medical professionals in understanding the rationale behind each predictive outcome.
[0013] Another objective of the present disclosure is to implement real-time monitoring of physiological parameters such as heart rate, blood pressure, and oxygen saturation by using wearable Internet of Things sensors and processing the data locally through edge computing infrastructure for immediate analysis and alerts.
[0014] Another objective of the present disclosure is to employ federated learning techniques to collaboratively train machine learning models across multiple decentralized healthcare facilities without transferring sensitive patient data, thereby maintaining data privacy and regulatory compliance.
[0015] Another objective of the present disclosure is to address the issue of class imbalance in stroke-related datasets by applying synthetic oversampling and algorithmic balancing methods to ensure equitable and accurate representation of stroke-prone cases in model training.
[0016] Another objective of the present disclosure is to design the prediction framework in a modular format that is supporting seamless integration with existing hospital record management systems and clinical decision-support tools currently being used in healthcare settings.
[0017] Another objective of the present disclosure is to enhance the generalizability and robustness of the prediction model by incorporating diverse patient datasets from different geographic, demographic, and clinical sources during model development and validation phases.
[0018] Another objective of the present disclosure is to minimize false-negative predictions in stroke diagnosis through optimized threshold tuning, advanced ensemble techniques, and continuous performance validation using real-world datasets across varied patient scenarios.
[0019] Yet another objective of the present disclosure is to generate real-time alerts and detailed risk factor summaries for physicians and caregivers by translating raw prediction outputs into interpretable dashboards with user-friendly formats suitable for clinical decision-making.
[0020] Yet another objective of the present disclosure is to improve overall patient outcomes by continuously refining stroke risk prediction models using feedback loops, historical data re-evaluation, and adaptive learning strategies that are adjusting to evolving medical knowledge and patient behaviour patterns.
[0021] In light of the above, in one aspect of the present disclosure, a brain stroke prediction system is disclosed herein. The system comprises a user device, the user device configured to continuously collect multimodal patient-specific data, the user device comprising. The system includes a wearable biosensor, operatively connected to the user device and configured to measure and transmit real-time physiological data including heart rate, blood pressure, body temperature, and oxygen saturation for stroke risk analysis. The system also includes a manual data entry interface, operatively connected to the user device and configured to capture user-entered lifestyle inputs, medical history, and symptomatic feedback. The system also includes a communication network, operatively connected to the user device, the communication network configured to facilitate bi-directional transmission of encrypted medical data, the communication network comprising Bluetooth, Wi-Fi, and hospital information system integration protocols for secure data exchange. The system also includes a plurality of sensors, operatively connected to the communication network and configured to detect and transmit internal and external conditions relevant to brain stroke prediction, the plurality of sensors comprising. The system also includes a pulse sensor, configured to continuously monitor pulse rate variability and detect abnormal cardiovascular fluctuations. The system also includes a temperature sensor, configured to monitor systemic temperature shifts that are indicative of febrile strokes. The system also includes a motion sensor, configured to detect sudden limb immobility, tremors, or postural changes associated with transient ischemic attacks. The system also includes an oxygen sensor, configured to monitor blood oxygen saturation and identify hypoxic risk factors. The system also includes a processing unit, operatively connected to the communication network and the user device and configured to perform real-time prediction of stroke risk using explainable hybrid artificial intelligence architecture, the processing unit comprising. The system also includes a data aggregation module configured to pre-process and synchronize physiological, historical, and behavioural data received from the user device and external sensors. The system also includes a hybrid deep learning module configured to implement convolution neural networks for processing imaging data and transformer-based networks for clinical and behavioural data to identify stroke patterns. The system also includes an explain ability module configured to execute interpretable models using Shapley additive explanations and local interpretable model-agnostic explanations for visualization of contributing factors. The system also includes a federated learning module configured to train and refine model weights across decentralized hospital nodes using encrypted gradient exchange without transmitting raw patient data. The system also includes a real-time inference engine module configured to perform stroke prediction at the edge level directly on the user device using lightweight neural model compression. The system also includes a user interface, operatively connected to the processing unit and configured to display stroke risk probabilities, explanatory visual outputs, and medical alerts to a healthcare provider or patient in real time, wherein the user interface is further configured to receive corrective inputs, annotations, and medical overrides from clinicians. The system also includes a display unit, operatively connected to the user interface, the display unit configured to show visual summaries of stroke risk trends, prediction heatmaps, and time-series alerts. The system also includes a storage unit, operatively connected to the processing unit, the storage unit configured to securely store historical patient records, intermediate model states, and risk assessment logs in an encrypted and retrievable format for longitudinal analysis and model retraining.
[0022] In one embodiment, the wearable biosensor is further configured to be operatively connected to the data aggregation module for transmitting real-time heart rate, blood pressure, temperature, and oxygen saturation data in a continuous manner for pre-processing and synchronization.
[0023] In one embodiment, the pulse sensor is further configured to be operatively connected to the hybrid deep learning module for transmitting pulse rate variability patterns to assist in the identification of cardiovascular irregularities contributing to stroke prediction.
[0024] In one embodiment, the motion sensor is further configured to be operatively connected to the data aggregation module and the explain ability module for transmitting postural and movement deviations to generate interpretable outputs for transient ischemic event detection.
[0025] In one embodiment, the federated learning module is further configured to be operatively connected to the real-time inference engine module for updating lightweight model parameters locally at the user device based on encrypted inter-hospital learning gradients.
[0026] In one embodiment, the explain ability module is further configured to be operatively connected to the user interface for visualizing stroke prediction contributing factors using Shapley additive values and interpretable model outputs.
[0027] In one embodiment, wherein the hybrid deep learning module is further configured to be operatively connected to the storage unit for storing updated model weights and prediction history associated with multimodal input vectors and neural inference outputs.
[0028] In one embodiment, the user interface is further configured to be operatively connected to the display unit for rendering stroke risk summaries, trend visualizations, and clinician-provided annotations received through the manual data entry interface.
[0029] In one embodiment, the temperature sensor is further configured to be operatively connected to the data aggregation module and the hybrid deep learning module for enabling stroke prediction under febrile physiological states.
[0030] In light of the above, in one aspect of the present disclosure, a method for brain stroke prediction system is disclosed herein. The method comprises initiating acquisition of physiological signals from a wearable biosensor. The method includes receiving user-entered lifestyle information, symptomatic feedback, and historical medical data through a manual data entry interface. The method also includes establishing secure bidirectional communication between the user device and a processing unit through a communication network. The method also includes acquiring external contextual data from a plurality of sensors, each sensor being operatively connected to the user device for continuous transmission of real-time biological and environmental metrics. The method also includes pre-processing and synchronizing all collected data using a data aggregation module. The method also includes implementing a hybrid deep learning architecture comprising a convolutional neural network for imaging-based analysis and a transformer-based network for behavioural and clinical data. The method also includes executing explainable artificial intelligence techniques using an explain ability module. The method also includes performing distributed training of neural weights across hospital-based nodes using a federated learning module and configured to exchange encrypted gradients while excluding raw patient data transmission. The method also includes generating stroke risk predictions in real time through a real-time inference engine module, the inference engine module being configured for edge-level execution on the user device. The method also includes displaying stroke risk levels, annotated visual outputs, and time-series indicators through a user interface. The method also includes storing historical stroke predictions, patient records, and intermediate inference states in a secure and retrievable format.
[0031] These and other advantages will be apparent from the present application of the embodiments described herein.
[0032] The preceding is a simplified summary to provide an understanding of some embodiments of the present invention. This summary is neither an extensive nor exhaustive overview of the present invention and its various embodiments. The summary presents selected concepts of the embodiments of the present invention in a simplified form as an introduction to the more detailed description presented below. As will be appreciated, other embodiments of the present invention are possible utilizing, alone or in combination, one or more of the features set forth above or described in detail below.
[0033] These elements, together with the other aspects of the present disclosure and various features are pointed out with particularity in the claims annexed hereto and form a part of the present disclosure. For a better understanding of the present disclosure, its operating advantages, and the specified object attained by its uses, reference should be made to the accompanying drawings and descriptive matter in which there are illustrated exemplary embodiments of the present disclosure.
BRIEF DESCRIPTION OF THE DRAWINGS
[0034] To describe the technical solutions in the embodiments of the present disclosure or in the prior art more clearly, the following briefly describes the accompanying drawings required for describing the embodiments or the prior art. Apparently, the accompanying drawings in the following description merely show some embodiments of the present disclosure, and a person of ordinary skill in the art can derive other implementations from these accompanying drawings without creative efforts. All of the embodiments or the implementations shall fall within the protection scope of the present disclosure.
[0035] The advantages and features of the present disclosure will become better understood with reference to the following detailed description taken in conjunction with the accompanying drawing, in which:
[0036] FIG. 1 illustrates a block diagram of a brain stroke prediction system, the system in accordance with an exemplary embodiment of the present disclosure;
[0037] FIG. 2 illustrates a flow chart of brain stroke prediction, in accordance with an exemplary embodiment of the present disclosure;
[0038] FIG. 3 illustrates a perspective view of a performance of proposed system, in accordance with an exemplary embodiment of the present disclosure;
[0039] FIG. 4 illustrates a perspective view of ROC and precision recall of proposed system, in accordance with an exemplary embodiment of the present disclosure.
[0040] Like reference, numerals refer to like parts throughout the description of several views of the drawing.
[0041] The brain stroke prediction system, the system and method thereof is illustrated in the accompanying drawings, which like reference letters indicate corresponding parts in the various figures. It should be noted that the accompanying figure is intended to present illustrations of exemplary embodiments of the present disclosure. This figure is not intended to limit the scope of the present disclosure. It should also be noted that the accompanying figure is not necessarily drawn to scale.
DETAILED DESCRIPTION OF THE DISCLOSURE
[0042] The following is a detailed description of embodiments of the disclosure depicted in the accompanying drawings. The embodiments are in such detail as to communicate the disclosure. However, the amount of detail offered is not intended to limit the anticipated variations of embodiments; on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the present disclosure.
[0043] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It may be apparent to one skilled in the art that embodiments of the present disclosure may be practiced without some of these specific details.
[0044] Various terms as used herein are shown below. To the extent a term is used, it should be given the broadest definition persons in the pertinent art have given that term as reflected in printed publications and issued patents at the time of filing.
[0045] The terms “a” and “an” herein do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced items.
[0046] The terms “having”, “comprising”, “including”, and variations thereof signify the presence of a component.
[0047] Referring now to FIG. 1 to FIG. 4 to describe various exemplary embodiments of the present disclosure. FIG. 1 illustrates a block diagram of a brain stroke prediction system, the system in accordance with an exemplary embodiment of the present disclosure.
[0048] The system 100 may include a user device 102, the user device 102 configured to continuously collect multimodal patient-specific data, the user device 102 comprising, a wearable biosensor 104, operatively connected to the user device 102 and configured to measure and transmit real-time physiological data including heart rate, blood pressure, body temperature, and oxygen saturation for stroke risk analysis, a manual data entry interface 106, operatively connected to the user device 102 and configured to capture user-entered lifestyle inputs, medical history, and symptomatic feedback, a communication network 108, operatively connected to the user device 102, the communication network 108 configured to facilitate bi-directional transmission of encrypted medical data, the communication network 108 comprising Bluetooth, Wi-Fi, and hospital information system integration protocols for secure data exchange, a plurality of sensors 110, operatively connected to the communication network 108 and configured to detect and transmit internal and external conditions relevant to brain stroke prediction, the plurality of sensors 110 comprising, a pulse sensor 112, configured to continuously monitor pulse rate variability and detect abnormal cardiovascular fluctuations, a temperature sensor 114, configured to monitor systemic temperature shifts that are indicative of febrile strokes, a motion sensor 116, configured to detect sudden limb immobility, tremors, or postural changes associated with transient ischemic attacks, an oxygen sensor 118, configured to monitor blood oxygen saturation and identify hypoxic risk factors, a processing unit 120, operatively connected to the communication network 108 and the user device 102 and configured to perform real-time prediction of stroke risk using an explainable hybrid artificial intelligence architecture, the processing unit 120 comprising, a data aggregation module 122 configured to pre-process and synchronize physiological, historical, and behavioural data received from the user device 102 and external sensors, a hybrid deep learning module 124 configured to implement convolution neural networks for processing imaging data and transformer-based networks for clinical and behavioural data to identify stroke patterns, an explain ability module 126 configured to execute interpretable models using Shapley additive explanations and local interpretable model-agnostic explanations for visualization of contributing factors, a federated learning module 128 configured to train and refine model weights across decentralized hospital nodes using encrypted gradient exchange without transmitting raw patient data;
a real-time inference engine module 130 configured to perform stroke prediction at the edge level directly on the user device 102 using lightweight neural model compression, a user interface 132, operatively connected to the processing unit 120 and configured to display stroke risk probabilities, explanatory visual outputs, and medical alerts to a healthcare provider or patient in real time, wherein the user interface 132 is further configured to receive corrective inputs, annotations, and medical overrides from clinicians, a display unit 134, operatively connected to the user interface 132, the display unit 134 configured to show visual summaries of stroke risk trends, prediction heatmaps, and time-series alerts, a storage unit 136, operatively connected to the processing unit 120, the storage unit 136 configured to securely store historical patient records, intermediate model states, and risk assessment logs in an encrypted and retrievable format for longitudinal analysis and model retraining.
[0049] The wearable biosensor 104 is further configured to be operatively connected to the data aggregation module 122 for transmitting real-time heart rate, blood pressure, temperature, and oxygen saturation data in a continuous manner for pre-processing and synchronization.
[0050] The pulse sensor 112 is further configured to be operatively connected to the hybrid deep learning module 124 for transmitting pulse rate variability patterns to assist in the identification of cardiovascular irregularities contributing to stroke prediction.
[0051] The motion sensor 116 is further configured to be operatively connected to the data aggregation module 122 and the explain ability module 126 for transmitting postural and movement deviations to generate interpretable outputs for transient ischemic event detection.
[0052] The federated learning module 128 is further configured to be operatively connected to the real-time inference engine module 130 for updating lightweight model parameters locally at the user device 102 based on encrypted inter-hospital learning gradients.
[0053] The explain ability module 126 is further configured to be operatively connected to the user interface 132 for visualizing stroke prediction contributing factors using Shapley additive values and interpretable model outputs.
[0054] The hybrid deep learning module 124 is further configured to be operatively connected to the storage unit 136 for storing updated model weights and prediction history associated with multimodal input vectors and neural inference outputs.
[0055] The user interface 132 is further configured to be operatively connected to the display unit 134 for rendering stroke risk summaries, trend visualizations, and clinician-provided annotations received through the manual data entry interface 106.
[0056] The temperature sensor 114 is further configured to be operatively connected to the data aggregation module 122 and the hybrid deep learning module 124 for enabling stroke prediction under febrile physiological states.
[0057] The method may include initiating acquisition of physiological signals from a wearable biosensor 104, receiving user-entered lifestyle information, symptomatic feedback, and historical medical data through a manual data entry interface 106, establishing secure bidirectional communication between the user device 102 and a processing unit 120 through a communication network 108, acquiring external contextual data from a plurality of sensors 110, each sensor being operatively connected to the user device 102 for continuous transmission of real-time biological and environmental metrics, pre-processing and synchronizing all collected data using a data aggregation module 122, implementing a hybrid deep learning architecture comprising a convolutional neural network for imaging-based analysis and a transformer-based network for behavioural and clinical data, executing explainable artificial intelligence techniques using an explain ability module 126, performing distributed training of neural weights across hospital-based nodes using a federated learning module 128 and configured to exchange encrypted gradients while excluding raw patient data transmission, generating stroke risk predictions in real time through a real-time inference engine module 130, the inference engine module being configured for edge-level execution on the user device 102, displaying stroke risk levels, annotated visual outputs, and time-series indicators through a user interface 132, storing historical stroke predictions, patient records, and intermediate inference states in a secure and retrievable format.
[0058] The user device 102 is continuously collecting multimodal patient-specific data by functioning as the central control unit of the brain stroke prediction system 100. The user device 102 receives inputs from both automated sensor feeds and manual data entries and maintains uninterrupted communication with other modules through the communication network 108. The user device 102 synchronizes data transmission and ensures real-time monitoring of physiological and lifestyle variables required for stroke risk analysis, offering seamless integration of wearable biosensor 104 data and manually captured health records to establish a continuous health profile. The user device 102 is operating consistently at step 102.
[0059] The wearable biosensor 104 is measuring and transmitting real-time physiological data that include heart rate, blood pressure, body temperature, and oxygen saturation. The wearable biosensor 104 is operatively connected to the user device 102 and is continuously collecting and updating vital parameters necessary for early identification of stroke-related abnormalities. This data is subsequently relayed to the processing unit 120 through the communication network 108 for risk evaluation. The wearable biosensor 104 ensures that no physiological anomaly goes unnoticed and contributes to uninterrupted health surveillance in the system. The wearable biosensor 104 is actively functioning at step 104.
[0060] The manual data entry interface 106 is capturing user-entered lifestyle information, symptomatic feedback, and medical history. The manual data entry interface 106 is enabling users and healthcare professionals to provide contextual information such as medication schedules, dietary habits, recent stress episodes, and pre-existing medical conditions. These manually entered inputs are transmitted to the processing unit 120 via the communication network 108 and are integrated with sensor-acquired data to enhance model accuracy and personalization. The manual data entry interface 106 is operating consistently and efficiently within the architecture at step 106.
[0061] The communication network 108 is facilitating bi-directional transmission of encrypted medical data between the user device 102, wearable biosensor 104, plurality of sensors 110, and processing unit 120. The communication network 108 is utilizing Bluetooth, Wi-Fi, and hospital information system integration protocols to ensure secure and uninterrupted data exchange. Real-time data from multiple sources is continuously flowing through the communication network 108, maintaining a coherent data pipeline essential for timely analysis and feedback delivery. The communication network 108 is functioning actively and integrally at step 108.
[0062] The plurality of sensors 110 is functioning as a composite unit comprising multiple specialized sensors configured to detect and transmit internal and external physiological conditions critical for stroke prediction. The plurality of sensors 110 includes a pulse sensor 112, temperature sensor 114, motion sensor 116, and oxygen sensor 118, all connected through the communication network 108. These sensors are continuously streaming multimodal data to the processing unit 120, enabling real-time aggregation and analysis. The plurality of sensors 110 is constantly contributing vital data at step 110.
[0063] The pulse sensor 112 is continuously monitoring pulse rate variability to detect abnormal cardiovascular fluctuations. The pulse sensor 112 is transmitting this data through the communication network 108 to the processing unit 120, where hybrid deep learning algorithms interpret the temporal patterns. Variations captured by the pulse sensor 112 play a critical role in identifying risk events such as arrhythmias or sudden drops in blood circulation associated with ischemic strokes. The pulse sensor 112 is constantly active and collecting cardiovascular indicators at step 112.
[0064] The temperature sensor 114 is recording systemic temperature fluctuations that signify febrile conditions often linked to inflammatory or haemorrhagic strokes. The temperature sensor 114 is working in real time and is integrated into the communication network 108, relaying temperature data for analysis and correlation with other symptoms. These readings are being factored into composite risk assessments by the hybrid deep learning module 124. The temperature sensor 114 is providing continuous thermal data to the processing pipeline at step 114.
[0065] The motion sensor 116 is detecting sudden changes in posture, limb immobility, or tremors that are indicative of transient ischemic attacks or early stroke events. The motion sensor 116 is capturing dynamic positional data and relaying this through the communication network 108 for temporal analysis by the hybrid deep learning module 124. These biomechanical changes serve as critical precursors to stroke risk. The motion sensor 116 is maintaining real-time feedback to the central system architecture at step 116.
[0066] The oxygen sensor 118 is monitoring blood oxygen saturation levels and identifying potential hypoxic conditions that contribute to stroke susceptibility. The oxygen sensor 118 is continuously streaming SpO2 readings through the communication network 108 to the processing unit 120, where it is analyzed alongside cardiovascular and neurological data. Persistent deviations are triggering alerts and contributing to predictive modelling. The oxygen sensor 118 is consistently operating at step 118.
[0067] The processing unit 120 is executing real-time prediction of stroke risk using an explainable hybrid artificial intelligence architecture. The processing unit 120 is receiving and analysing multimodal inputs from sensors, wearable biosensor 104, and manual data entry interface 106, integrating them using various analytical modules to generate probabilistic stroke assessments. The processing unit 120 is serving as the computational core of the system and is functioning actively at step 120.
[0068] The data aggregation module 122 is pre-processing and synchronizing incoming physiological, behavioural, and historical data. The data aggregation module 122 is harmonizing data formats and aligning time-series inputs to create a consistent and complete dataset for subsequent analysis. This pre-processing step is enhancing model accuracy and mitigating signal noise. The data aggregation module 122 is performing data normalization and synchronization continuously at step 122.
[0069] The hybrid deep learning module 124 is implementing convolutional neural networks for imaging data and transformer-based networks for textual and behavioural data to identify potential stroke patterns. The hybrid deep learning module 124 is operating across multiple data modalities and integrating insights to build a robust risk profile. These deep learning structures are dynamically adapting to input data and delivering high-fidelity predictions. The hybrid deep learning module 124 is executing cross-modal inference at step 124.
[0070] The explain ability module 126 is employing Shapley additive explanations and local interpretable model-agnostic explanations to visualize and interpret the contributions of each input variable to stroke risk predictions. The explain ability module 126 is generating heatmaps, impact scores, and annotated visual outputs for clinicians and patients. This transparency is enhancing trust and enabling informed decision-making. The explain ability module 126 is actively translating complex model outputs into interpretable insights at step 126.
[0071] The federated learning module 128 is training and refining neural model weights across decentralized hospital nodes using encrypted gradient exchange protocols. The federated learning module 128 is preserving patient privacy by avoiding raw data transmission while collaboratively enhancing model performance. This distributed learning paradigm is ensuring system-wide robustness and adaptability. The federated learning module 128 is continuously managing decentralized updates at step 128.
[0072] The real-time inference engine module 130 is performing stroke prediction at the edge level using compressed neural models embedded within the user device 102. The real-time inference engine module 130 is delivering low-latency predictions and enabling on-device intelligence without reliance on remote servers. This module is ensuring uninterrupted functionality even in limited connectivity environments. The real-time inference engine module 130 is generating instant risk outputs at step 130.
[0073] The user interface 132 is displaying stroke risk probabilities, explanatory visualizations, and medical alerts in real time. The user interface 132 is enabling patients and clinicians to view health metrics, receive early warnings, and annotate data. It is also accepting medical overrides and corrective feedback from healthcare providers. The user interface 132 is providing bi-directional engagement with the system at step 132.
[0074] The display unit 134 is presenting visual summaries of stroke prediction trends, including risk graphs, heatmaps, and alert timelines. The display unit 134 is operatively connected to the user interface 132 and is enhancing information accessibility for users. Clinicians are using this visual interface to track longitudinal trends and assess intervention needs. The display unit 134 is generating real-time visuals at step 134.
[0075] The storage unit 136 is securely archiving historical health data, intermediate model states, and risk assessments. The storage unit 136 is operating with encryption protocols to ensure data privacy and regulatory compliance. This archived data is serving as a resource for longitudinal analysis, trend monitoring, and periodic model retraining. The storage unit 136 is maintaining persistent records at step 136.
[0076] FIG. 2 illustrates a flow chart of brain stroke prediction, in accordance with an exemplary embodiment of the present disclosure.
[0077] At 202, initiate acquisition of physiological signals from a wearable biosensor.
[0078] At 204, receive user-entered lifestyle information, symptomatic feedback, and historical medical data through a manual data entry interface.
[0079] At 206, establish secure bidirectional communication between the user device and a processing unit through a communication network.
[0080] At 208, acquire external contextual data from a plurality of sensors, each sensor being operatively connected to the user device for continuous transmission of real-time biological and environmental metrics.
[0081] At 210, preprocess and synchronizing all collected data using a data aggregation module.
[0082] At 212, implement a hybrid deep learning architecture comprising a convolutional neural network for imaging-based analysis and a transformer-based network for behavioural and clinical data.
[0083] At 214, execute explainable artificial intelligence techniques using an explain ability module.
[0084] At 216, perform distributed training of neural weights across hospital-based nodes using a federated learning module and configured to exchange encrypted gradients while excluding raw patient data transmission.
[0085] At 218, generate stroke risk predictions in real time through a real-time inference engine module, the inference engine module being configured for edge-level execution on the user device.
[0086] At 220, display stroke risk levels, annotated visual outputs, and time-series indicators through a user interface.
[0087] At 222, store historical stroke predictions, patient records, and intermediate inference states in a secure and retrievable format.
[0088] FIG. 3 illustrates a perspective view of a performance of proposed system, in accordance with an exemplary embodiment of the present disclosure.
[0089] FIG. 3 illustrates a perspective view of the performance evaluation metrics for the proposed Boost-based system, in accordance with an exemplary embodiment of the present disclosure. The bar graph demonstrates high values across key performance indicators—accuracy, precision, recall, and F1 score—indicating the system’s strong classification capability and robustness in real-world application.
[0090] FIG. 4 illustrates a perspective view of ROC and precision recall of proposed system, in accordance with an exemplary embodiment of the present disclosure.
[0091] FIG. 4 illustrates a perspective view of the ROC and Precision-Recall curves for the proposed system, in accordance with an exemplary embodiment of the present disclosure. The Boost model achieved the highest AUC of 0.99, followed by Random Forest 0.95, ANN 0.91, and SVM 0.93, demonstrating superior performance in classification tasks with minimal false positives.
[0092] While the invention has been described in connection with what is presently considered to be the most practical and various embodiments, it will be understood that the invention is not to be limited to the disclosed embodiments, but on the contrary, is intended to cover various modifications and equivalent arrangements included within the scope of the appended claims.
[0093] A person of ordinary skill in the art may be aware that, in combination with the examples described in the embodiments disclosed in this specification, units and algorithm steps may be implemented by electronic hardware, computer software, or a combination thereof.
[0094] The foregoing descriptions of specific embodiments of the present disclosure have been presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the present disclosure to the precise forms disclosed, and many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described to best explain the principles of the present disclosure and its practical application, and to thereby enable others skilled in the art to best utilize the present disclosure and various embodiments with various modifications as are suited to the particular use contemplated. It is understood that various omissions and substitutions of equivalents are contemplated as circumstances may suggest or render expedient, but such omissions and substitutions are intended to cover the application or implementation without departing from the scope of the present disclosure.
[0095] Disjunctive language such as the phrase “at least one of X, Y, Z,” unless specifically stated otherwise, is otherwise understood with the context as used in general to present that an item, term, etc., may be either X, Y, or Z, or any combination thereof (e.g., X, Y, and/or Z). Thus, such disjunctive language is not generally intended to, and should not, imply that certain embodiments require at least one of X, at least one of Y, or at least one of Z to each be present.
[0096] In a case that no conflict occurs, the embodiments in the present disclosure and the features in the embodiments may be mutually combined. The foregoing descriptions are merely specific implementations of the present disclosure, but are not intended to limit the protection scope of the present disclosure. Any variation or replacement readily figured out by a person skilled in the art within the technical scope disclosed in the present disclosure shall fall within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure shall be subject to the protection scope of the claims.
, Claims:I/We Claim:
1. A brain stroke prediction system (100), the system (100) comprising:
a user device (102), the user device (102) configured to continuously collect multimodal patient-specific data, the user device (102) comprising:
a wearable biosensor (104), operatively connected to the user device (102) and configured to measure and transmit real-time physiological data including heart rate, blood pressure, body temperature, and oxygen saturation for stroke risk analysis;
a manual data entry interface (106), operatively connected to the user device (102) and configured to capture user-entered lifestyle inputs, medical history, and symptomatic feedback;
a communication network (108), operatively connected to the user device (102), the communication network (108) configured to facilitate bi-directional transmission of encrypted medical data, the communication network (108) comprising Bluetooth, Wi-Fi, and hospital information system integration protocols for secure data exchange;
a plurality of sensors (110), operatively connected to the communication network (108) and configured to detect and transmit internal and external conditions relevant to brain stroke prediction, the plurality of sensors (110) comprising:
a pulse sensor (112), configured to continuously monitor pulse rate variability and detect abnormal cardiovascular fluctuations;
a temperature sensor (114), configured to monitor systemic temperature shifts that are indicative of febrile strokes;
a motion sensor (116), configured to detect sudden limb immobility, tremors, or postural changes associated with transient ischemic attacks;
an oxygen sensor (118), configured to monitor blood oxygen saturation and identify hypoxic risk factors;
a processing unit (120), operatively connected to the communication network (108) and the user device (102) and configured to perform real-time prediction of stroke risk using an explainable hybrid artificial intelligence architecture, the processing unit (120) comprising:
a data aggregation module (122) configured to pre-process and synchronize physiological, historical, and behavioural data received from the user device (102) and external sensors;
a hybrid deep learning module (124) configured to implement convolution neural networks for processing imaging data and transformer-based networks for clinical and behavioural data to identify stroke patterns;
an explain ability module (126) configured to execute interpretable models using Shapley additive explanations and local interpretable model-agnostic explanations for visualization of contributing factors;
a federated learning module (128) configured to train and refine model weights across decentralized hospital nodes using encrypted gradient exchange without transmitting raw patient data;
a real-time inference engine module (130) configured to perform stroke prediction at the edge level directly on the user device (102) using lightweight neural model compression;
a user interface (132), operatively connected to the processing unit (120) and configured to display stroke risk probabilities, explanatory visual outputs, and medical alerts to a healthcare provider or patient in real time, wherein the user interface (132) is further configured to receive corrective inputs, annotations, and medical overrides from clinicians.
a display unit (134), operatively connected to the user interface (132), the display unit (134) configured to show visual summaries of stroke risk trends, prediction heatmaps, and time-series alerts;
a storage unit (136), operatively connected to the processing unit (120), the storage unit (136) configured to securely store historical patient records, intermediate model states, and risk assessment logs in an encrypted and retrievable format for longitudinal analysis and model retraining.
2. The system (100) as claimed in claim 1, wherein the wearable biosensor (104) is further configured to be operatively connected to the data aggregation module (120) for transmitting real-time heart rate, blood pressure, temperature, and oxygen saturation data in a continuous manner for pre-processing and synchronization.
3. The system (100) as claimed in claim 1, wherein the pulse sensor (112) is further configured to be operatively connected to the hybrid deep learning module (122) for transmitting pulse rate variability patterns to assist in the identification of cardiovascular irregularities contributing to stroke prediction
4. The system (100) as claimed in claim 1, wherein the motion sensor (116) is further configured to be operatively connected to the data aggregation module (120) and the explain ability module (124) for transmitting postural and movement deviations to generate interpretable outputs for transient ischemic event detection.
5. The system (100) as claimed in claim 1, wherein the federated learning module (126) is further configured to be operatively connected to the real-time inference engine module (128) for updating lightweight model parameters locally at the user device (102) based on encrypted inter-hospital learning gradients.
6. The system (100) claimed in claim 1, wherein the explain ability module (124) is further configured to be operatively connected to the user interface (130) for visualizing stroke prediction contributing factors using Shapley additive values and interpretable model outputs.
7. The system (100) as claimed in claim 1, wherein the hybrid deep learning module (122) is further configured to be operatively connected to the storage unit (134) for storing updated model weights and prediction history associated with multimodal input vectors and neural inference outputs.
8. The system (100) as claimed in claim 1, wherein the user interface (130) is further configured to be operatively connected to the display unit (132) for rendering stroke risk summaries, trend visualizations, and clinician-provided annotations received through the manual data entry interface (106).
9. The system (100) as claimed in claim 1, wherein the temperature sensor (114) is further configured to be operatively connected to the data aggregation module (120) and the hybrid deep learning module (122) for enabling stroke prediction under febrile physiological states.
10. A method (100) for brain stroke prediction, the method (100) comprising:
initiating acquisition of physiological signals from a wearable biosensor (104);
receiving user-entered lifestyle information, symptomatic feedback, and historical medical data through a manual data entry interface (106);
establishing secure bidirectional communication between the user device (102) and a processing unit (120) through a communication network (108);
acquiring external contextual data from a plurality of sensors (110), each sensor being operatively connected to the user device (102) for continuous transmission of real-time biological and environmental metrics;
pre-processing and synchronizing all collected data using a data aggregation module (120);
implementing a hybrid deep learning architecture comprising a convolutional neural network for imaging-based analysis and a transformer-based network for behavioural and clinical data;
executing explainable artificial intelligence techniques using an explain ability module (124);
performing distributed training of neural weights across hospital-based nodes using a federated learning module (126) and configured to exchange encrypted gradients while excluding raw patient data transmission;
generating stroke risk predictions in real time through a real-time inference engine module (128), the inference engine module being configured for edge-level execution on the user device (102);
displaying stroke risk levels, annotated visual outputs, and time-series indicators through a user interface (130);
storing historical stroke predictions, patient records, and intermediate inference states in a secure and retrievable format.