Ai-driven real-time healthcare management system with multimodal data

Abstract

A system for real-time healthcare management comprises a data processing unit configured to handle multimodal data inputs, including a convolutional neural network (CNN) for image classification and a large language model (LLM) for generating natural language responses. The system integrates a biometric data collection module for receiving data from wearable devices, and an integration module that facilitates communication between the CNN and LLM as well as external healthcare systems. A user interface allows for the input of images and queries, and displays processed data. The system includes a data storage unit for storing health data and a communication module for secure data exchange. This system enables comprehensive healthcare management by processing visual and textual data, integrating real-time biometric monitoring, and facilitating interaction with external healthcare providers.

Description

Title: Al-Driven Real-Time Healthcare Management System with Multimodal Data Integration and Biometric Monitoring.Field of Invention.

[0001] The present invention relates generally to systems and methods for healthcare management, specifically involving the integration of visual and textual data processing with biometric monitoring to enhance patient care and operational efficiency.Background

[0002] Current healthcare systems often face significant challenges in delivering efficient and accurate medical services. Traditional methods for medication management and patient care typically rely on manual processes and isolated data inputs, leading to inefficiencies and a higher risk of errors. Existing systems often lack the capability to integrate multimodal data inputs, such as textual and visual data, which are crucial for comprehensive healthcare management. Moreover, real-time predictive analytics and personalized medical advice based on continuous monitoring of biometric data are not widely available, resulting in reactive rather than proactive health management.

[0003] Conventional medication identification methods depend heavily on manual verification, which can be time-consuming and prone to errors, especially in distinguishing between medications with similar appearances. The inability to accurately identify medications can lead to significant medication errors, adversely affecting patient safety and treatment outcomes. Additionally, current systems do not effectively integrate real-time data from wearable devices to monitor health metrics continuously, thereby missing opportunities for early detection of potential health issues.

[0004] Another drawback in current healthcare systems is the lack of seamless interaction between different Al models that process diverse data types. Typically, systems do not efficiently combine visual data analysis (such as pill identification) with natural language processing to generate detailed and accurate medical advice. This gap hinders the ability to provide holistic and informed medical guidance based on integrated data analysis.

[0005] Furthermore, existing systems often struggle with providing dynamic and interactive patient engagement tools that can offer timely and personalized healthcare recommendations. Many systems lack advanced interfaces that facilitate easy interaction and data visualization for patients, reducing their engagement and adherence to prescribed health regimens.

[0006] The limitations in existing healthcare platforms highlight the need for an advanced system that can leverage cutting-edge Al technologies to address these challenges. It is within this context that the present invention is provided.

[0007] The integration of a Large Language Model (LLM) and a Convolutional Neural Network (CNN) offers a promising solution by enabling comprehensive data analysis and real-time predictive capabilities. Such a system can significantly enhance medication identification accuracy, provide real-time health monitoring, and deliver personalized medical advice, thereby improving patient outcomes and operational efficiencies in healthcare management.Summary

[0008] The present invention provides a system for healthcare management, comprising a data processing unit configured to receive and process multimodal data inputs. The data processing unit includes a visual data processing module with a convolutional neural network (CNN) for processing and classifying images of medications and a natural language processing module with a large language model (LLM) for generating natural language responses. The system further includes a biometric data collection module for real-time biometric data from wearable devices, an integration module to facilitate communication between the visual data processing module and the natural language processing module as well as external healthcare systems, a user interface for receiving user inputs and displaying processed data, a data storage unit for storing user-specific health data, and a communication module for data transmission between the system and external devices or networks.

[0009] In some embodiments, the visual data processing module comprises multiple convolutional layers for feature extraction, activation functions to introduce nonlinearity, pooling layers to reduce spatial dimensions and control overfitting, a flattening layer to convert outputs into a one-dimensional array, and fully connected dense layers for classification of medication images. This architecture ensures accurate and efficient image processing.

[0010] In further embodiments, the natural language processing module comprises a transformer-based architecture utilizing multi-head self-attention mechanisms, positional encodings, feed-forward neural networks, a pre-training phase on general and specialized medical datasets, and a fine-tuning phase for optimizing medical tasks. This enhances the system’s ability to provide accurate and contextually relevant responses.

[0011] In yet further embodiments, the biometric data collection module integrates with various wearable devices, including smartwatches, fitness trackers, smart clothing, medical-grade wearables, and embedded devices. This allows comprehensive realtime health monitoring.

[0012] In some embodiments, the integration module includes an API gateway for managing API requests, a middleware service for formatting data, and mechanisms for secure communication compliant with regulations such as HIPAA and GDPR. This ensures seamless and secure data exchange.

[0013] In further embodiments, the user interface features a responsive web and mobile application, capabilities for capturing user inputs, displays for visualizing processed data and health insights, and notification mechanisms for alerts based on predictive analysis. This enhances user interaction and engagement.

[0014] In yet further embodiments, the data storage unit includes secure and scalable databases, encryption protocols for data security, and logging and monitoring systems for compliance with regulatory requirements. This provides robust data management.

[0015] In some embodiments, the visual data processing module performs additional functionalities, including counterfeit pill detection, inventory management, and pharmaceutical research support. This broadens the module's application beyond basic image classification.

[0016] In further embodiments, the natural language processing module generates detailed health reports, recommends healthcare professionals and facilities, schedules and manages healthcare appointments, integrates with external healthcare provider systems, and tracks medication adherence, providing dosage reminders based on prescription data. These capabilities enhance the system's utility in healthcare management.

[0017] In yet further embodiments, the system includes predictive analytics capabilities for health risk assessment, disease progression monitoring, and early warning systems. It also features a feedback mechanism for users to report on the accuracy and usefulness of predictive insights, allowing continuous improvement of Al models, and a user profile analysis module to tailor advice based on individual patient profiles, considering medical history, allergies, and current medications.

[0018] In some embodiments, the system utilizes data augmentation techniques during model training to improve robustness to variations in input data, implements a microservices architecture for modular development and deployment, employs continuous integration and continuous deployment practices for rapid iteration and high-quality software production, and ensures compliance with healthcare regulations and data privacy laws through comprehensive data governance policies.

[0019] In further embodiments, the visual data processing module includes an image preprocessing sub-module to resize images to a uniform dimension and apply normalization techniques, and data augmentation functionalities such as rotations, flipping, and color adjustments to simulate various real-world conditions and improve model robustness.

[0020] In yet further embodiments, the natural language processing module includes a context management sub-module to maintain state and context of user interactions, ensuring continuity and relevance in generated responses, and a dialoguemanagement sub-module to handle multi-turn conversations, enabling the system to provide coherent and contextually appropriate responses over extended interactions.

[0021] In some embodiments, the biometric data collection module includes interfaces for integrating with home monitoring devices such as connected scales, sleep monitoring systems, and home ECG monitors, and capabilities to calibrate and validate data from various sensors to ensure accuracy and reliability of biometric measurements.

[0022] In further embodiments, the integration module includes an authentication and authorization sub-module to manage user access controls and ensure secure data exchange with external systems, and a data transformation sub-module to convert and normalize data formats from various sources for consistent processing and analysis.

[0023] In yet further embodiments, the user interface includes interactive elements such as body maps for symptom input and visualization tools for displaying health trends and insights, and configurable notification settings allowing users to personalize alerts and reminders based on their preferences.

[0024] In some embodiments, the data storage unit includes capabilities for managing historical data archives, allowing users to access and review past health data and interactions, and backup and disaster recovery mechanisms to ensure data integrity and availability in case of system failures.

[0025] In further embodiments, the visual data processing module generates alerts for potential medication errors by cross-referencing pill images with user-specific prescription data, and provides educational information about identified medications, including usage instructions and potential side effects.

[0026] In yet further embodiments, the system comprises a patient engagement module configured to provide personalized health tips, wellness programs, and motivational messages based on user data, and a healthcare provider interfaceallowing doctors to review patient data, provide consultations, and update treatment plans directly within the system.

[0027] In some embodiments, the biometric data collection module includes real-time synchronization capabilities with external health data repositories, ensuring that the latest biometric data is always available for analysis, and anomaly detection algorithms to flag irregular biometric readings for further investigation by healthcare professionals.Brief Description of the Drawings

[0028] Various embodiments of the invention are disclosed in the following detailed description and accompanying drawings.

[0029] FIG. 1A illustrates an example overall system architecture for the real-time healthcare management system.

[0030] FIG.1 B illustrates an example diagram of the software modules implemented by the one or more servers of the healthcare management system.

[0031] FIG. 2A illustrates an example entity-relationship diagram showing the system architecture and interactions between different entities.

[0032] FIG. 2B illustrates an example design class diagram detailing the relationships and interactions between various classes within the System.

[0033] FIG. 3 illustrates an example flow diagram of a generalized process for a patient accessing a service of the system and the interaction between a CNN and LLM.

[0034] FIG. 4 illustrates an example sequence diagram for medical adherence, showing the workflow involving the System to ensure patients adhere to their medication schedules.

[0035] FIG. 5A illustrates an example sequence diagram for patient requesting pill identification, detailing the interaction between the patient, the System, the CNN model, and the LLM model.

[0036] FIG. 5B illustrates an example sequence diagram for health screening, outlining the interaction between the patient, the doctor, the hospital system, the System, and the LLM model.

[0037] FIG. 50 illustrates an example sequence diagram for alerts and notifications, showing how the System monitors biometric data, medication adherence, and daily calorie count to generate and send relevant alerts.

[0038] FIG. 5D illustrates an example sequence diagram for predictive analytics, demonstrating how the System uses biometric data to generate predictive health alerts and recommend healthcare professional visits if necessary.

[0039] Common reference numerals are used throughout the figures and the detailed description to indicate like elements. One skilled in the art will readily recognize that the above figures are examples and that other architectures, modes of operation, orders of operation, and elements / functions can be provided and implemented without departing from the characteristics and features of the invention, as set forth in the claims.Detailed Description and Preferred Embodiment

[0040] The following is a detailed description of exemplary embodiments to illustrate the principles of the invention. The embodiments are provided to illustrate aspects of the invention, but the invention is not limited to any embodiment. The scope of the invention encompasses numerous alternatives, modifications and equivalent; it is limited only by the claims.

[0041] Numerous specific details are set forth in the following description in order to provide a thorough understanding of the invention. However, the invention may be practiced according to the claims without some or all of these specific details. For the purpose of clarity, technical material that is known in the technical fields related to theinvention has not been described in detail so that the invention is not unnecessarily obscured.DEFINITIONS:

[0042] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0043] As used herein, the term “and / or” includes any combinations of one or more of the associated listed items.

[0044] As used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well as the singular forms, unless the context clearly indicates otherwise.

[0045] It will be further understood that the terms “comprises” and / or “comprising,” when used in this specification, specify the presence of stated features, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0046] The terms "first," "second," and the like are used to distinguish different elements or features, but these elements or features should not be limited by these terms. A first element or feature described can be referred to as a second element or feature and vice versa without departing from the teachings of the present disclosure.

[0047] The term "data processing unit" refers to any computational component configured to receive, process, and analyze multimodal data inputs. This includes, but is not limited to, integrated circuits, microprocessors, and software modules capable of executing algorithms for image recognition and natural language processing. In one example implementation, the data processing unit comprises a high-performance microprocessor running TensorFlow and Keras frameworks for CNN operations, coupled with a transformer-based LLM model deployed on a dedicated server.

[0048] The term "visual data processing module" refers to a component of the data processing unit that handles the analysis and classification of image data. This includes, but is not limited to, convolutional neural networks (CNNs) with layers forconvolution, pooling, and fully connected operations. In one example implementation, the visual data processing module uses a CNN with multiple Conv2D layers, ReLU activation functions, and max-pooling layers, implemented using TensorFlow and Keras, to identify and classify pill images captured by a user's mobile device.

[0049] The term "natural language processing module" refers to a component of the data processing unit that processes and generates text-based data. This includes, but is not limited to, large language models (LLMs) based on transformer architectures utilizing self-attention mechanisms. In one example implementation, the natural language processing module employs a pre-trained LLM like Llama 3, fine-tuned on medical literature and patient interaction data to provide accurate medical advice and responses.

[0050] The term "biometric data collection module" refers to any device or software component that collects real-time physiological data from users. This includes, but is not limited to, sensors in wearable devices such as smartwatches, fitness trackers, and medical-grade monitors. In one example implementation, the biometric data collection module interfaces with Apple Watch through HealthKit to gather data such as heart rate, blood oxygen levels, and sleep patterns, which are then processed for health monitoring and predictive analytics.

[0051] The term "integration module" refers to any software or hardware component that facilitates communication between different subsystems of the data processing unit and external systems. This includes, but is not limited to, API gateways, middleware services, and data transformation layers. In one example implementation, the integration module employs RESTful APIs to transmit data between the CNN and LLM, ensuring structured data formats and secure communication protocols are maintained.

[0052] The term "user interface" refers to any software or hardware component that allows a user to interact with the system. This includes, but is not limited to, graphical user interfaces (GUIs) on web browsers, mobile applications, and other display devices. In one example implementation, the user interface is a mobile application that allows users to capture pill images, input symptoms, and receive real-time medical advice, with intuitive design elements for easy navigation and interaction.

[0053] The term "data storage unit" refers to any component or system used to store data persistently. This includes, but is not limited to, databases, cloud storage services, and local storage devices. In one example implementation, the data storage unit uses a scalable cloud-based database system like PostgreSQL, with encryption for data at rest and in transit, to securely store user health data, interaction history, and system-generated reports.

[0054] The term "communication module" refers to any component or system used to facilitate data exchange between the system and external networks or devices. This includes, but is not limited to, network interfaces, wireless communication protocols, and data encryption technologies. In one example implementation, the communication module uses SSL / TLS protocols to ensure secure data transmission between the system and external healthcare providers, supporting various communication standards such as HTTP, MQTT, and WebSockets.

[0055] The term "convolutional neural network" (CNN) refers to a class of deep learning algorithms used primarily for processing image data. This includes, but is not limited to, architectures with convolutional layers, pooling layers, and fully connected layers. The machine learning models for image processing of medical pills that could be used also include Visual Transformer (ViT), ResNet, VGG (Visual Geometry Group), MobileNetV2, Swin Transformer, and DenseNet. In one example implementation, the CNN is designed with a series of Conv2D layers that extract features from input images, max-pooling layers to reduce dimensionality, and dense layers to classify the images based on learned features.

[0056] The term "large language model" (LLM) refers to a type of artificial intelligence model designed to understand and generate human language. This includes, but is not limited to, transformer-based architectures that utilize multi-head self-attention mechanisms. In one example implementation, the LLM is pre-trained on a diverse corpus of general and medical texts, and then fine-tuned with specific medical dialogues and case studies to accurately interpret and respond to health-related queries.

[0057] The term "wearable devices" refers to electronic devices worn on the body that collect and transmit physiological data. This includes, but is not limited to,smartwatches, fitness bands, and medical-grade monitors. In one example implementation, wearable devices such as Fitbit or Apple Watch are used to continuously monitor metrics like heart rate and sleep patterns, which are then transmitted to the system for further analysis and health monitoring.DESCRIPTION OF DRAWINGS

[0058] The present invention relates to a system and method for real-time healthcare management that integrates multimodal data processing, including visual and textual data, with continuous biometric monitoring. This system is designed to address several shortcomings of existing healthcare management systems, which often rely on manual processes, isolated data inputs, and lack the capability to provide real-time predictive analytics and personalized medical advice.

[0059] Traditional healthcare systems typically face challenges such as inefficient medication identification, limited integration of real-time biometric data, and a lack of seamless interaction between different types of data processing models. These limitations can lead to medication errors, delayed detection of health issues, and fragmented patient care. The present invention overcomes these challenges by leveraging advanced artificial intelligence (Al) technologies, specifically a convolutional neural network (CNN) for image recognition and a large language model (LLM) for natural language processing, to provide a comprehensive healthcare management solution.

[0060] The system comprises a data processing unit that receives and processes multimodal data inputs. The CNN within the visual data processing module is responsible for accurately identifying and classifying images of medications, thereby reducing the risk of medication errors and enhancing patient safety. The LLM within the natural language processing module generates detailed and contextually relevant medical responses, improving the quality of medical advice provided to patients.

[0061] A biometric data collection module integrates data from wearable devices, enabling continuous monitoring of vital health metrics such as heart rate, blood pressure, and glucose levels. This real-time data is crucial for early detection of potential health issues and proactive health management. The integration module facilitates seamless communication between the visual and textual data processingcomponents and external healthcare systems, ensuring a cohesive and efficient flow of information.

[0062] The user interface is designed to be intuitive and user-friendly, allowing patients to easily input symptoms, capture pill images, and receive real-time medical advice. The system also includes a data storage unit that securely stores user health data, interaction history, and system-generated reports, ensuring that all relevant information is readily accessible and protected.

[0063] By integrating multimodal data processing, continuous biometric monitoring, and advanced Al-driven analytics, the present invention provides a robust and efficient healthcare management system. This system not only addresses the inefficiencies and limitations of traditional methods but also enhances patient care through accurate medication identification, real-time health monitoring, and personalized medical guidance.

[0064] Referring now to the drawings, FIG. 1A illustrates a block diagram of an example system architecture for the real-time healthcare management system, which integrates multimodal data processing, continuous biometric monitoring, and advanced Al-driven analytics to enhance patient care and operational efficiency.

[0065] At the core of the system is a server 100 that hosts the main functionalities and modules essential for the operation of the healthcare management system. The server 100 is linked to a set of databases 102 that securely store user health data, interaction histories, system-generated reports, and other relevant information. These databases 102 are designed with robust encryption protocols to ensure data security and compliance with healthcare regulations such as HI PAA and GDPR.

[0066] Users 110 can interact with the system in various ways. In the illustrated example, a patient 104 interacts with the system using a first user device 106, which is a smartphone. The patient 104 accesses the system via a dedicated app 108 installed on the first user device 106. This app 108 features an intuitive user interface that enables the patient 104 to input symptoms, capture images of medications, and receive real-time medical advice. The app 108 is designed to be user-friendly, providing easy navigation and access to various system functionalities.

[0067] A doctor 112 uses a second user device 114, which in the example is a tablet, to interact with the system. The doctor 112 can access patient data, review health reports, provide medical consultations, and update treatment plans through the app 108. The interface for the doctor 112 is tailored to support clinical workflows, ensuring efficient management of patient interactions and treatment documentation.

[0068] A system administrator 116 utilizes a third user device 118, which in the example is a laptop, to manage system operations. The system administrator 116 is responsible for user access management, system maintenance, data security, and compliance monitoring. The administrator interface is designed to provide comprehensive control over system settings and user permissions, ensuring that the system operates smoothly and securely.

[0069] Referring to FIG.1 B, a simplified diagram is shown of the software modules implemented by the one or more servers.

[0070] The cloud server 100 hosts several modules, including the visual data processing module 120 and the natural language processing module 122. The visual data processing module 120 incorporates and a convolutional neural network (CNN) 124, which processes and classifies images of medications captured by the patient's user device 106. The CNN 124 is designed with multiple convolutional layers, pooling layers, and fully connected dense layers, using TensorFlow and Keras frameworks. This configuration allows the CNN 124 to accurately identify medications by analyzing features such as shape, color, and markings.

[0071] The natural language processing module 122 includes a large language model (LLM) 126 that processes natural language inputs and generates contextually relevant responses. The LLM 126 is based on a transformer architecture that utilizes multihead self-attention mechanisms and positional encodings. It is pre-trained on a diverse corpus of general and medical texts and fine-tuned with specific medical dialogues and case studies to optimize its performance in healthcare-related interactions.

[0072] The interactions between the CNN 124 and LLM 126 are what, in combination with the various peripheral devices from which data is received, facilitate the majority of the system’s core functionalities.

[0073] The system also includes a biometric data collection module 128, which integrates with various wearable devices 130 used by the patient 104. These devices, such as smartwatches, fitness trackers, and medical-grade monitors, collect real-time biometric data including heart rate, blood pressure, glucose levels, and sleep patterns. The biometric data collection module 128 ensures continuous monitoring and analysis of these health metrics, enabling early detection of potential health issues and proactive management of the patient's health.

[0074] An integration module 132 facilitates seamless communication between the visual data processing module 120, the natural language processing module 122, and external healthcare systems 134. This module employs secure API gateways 136 to manage data exchange, ensuring that the system can interact with third-party healthcare providers, medication databases, and other relevant systems. The integration module 132 supports the provision of detailed analysis and insights derived from the system to these third-party entities, enhancing the overall healthcare ecosystem.

[0075] The system's user interface 110 allows for the input of various data types, including images and natural language queries, and displays processed data and generated responses. The interface is designed to support patient engagement and provide real-time insights, alerts, and recommendations based on the continuous analysis performed by the system.

[0076] The data storage unit 102 is a critical component that ensures the secure storage of all data processed by the system. It employs advanced encryption techniques for data at rest and in transit, and includes logging and monitoring systems to track data access and system events, ensuring compliance with healthcare regulations and maintaining data integrity.

[0077] FIG. 2A illustrates an Entity-Relationship Diagram (ERD) for the system architecture, showcasing the various entities involved in the System and their interactions. The SystemAdministrator profile 202 manages the Patient profile 204 and the Integrationsystem 206. The SystemAdministrator 202 is responsible for managing user access and performing system maintenance, ensuring the system operates smoothly and securely.

[0078] The Patient profile 204 inherits from the User entity 208, which includes general user attributes such as UserID, Name, Email, PhoneNumber, and Password. The Patient profile 204 includes additional functionalities specific to patient interactions with the system, such as login(), capturePilllmage(), provideBiometricData(), requestHealthScreening(), and confirmAppointmentBooking().

[0079] The Integrationsystem 206 represents external systems with attributes such as SystemID, SystemName, and SystemType. It includes the operation authorizeDataAccess(), which facilitates secure communication and data exchange between the System and external healthcare systems.

[0080] The Doctor entity 210, also inheriting from the User entity 208, includes attributes specific to healthcare providers, such as DoctorlD and Specialty. The Doctor 210 interacts with the Appointment entity 212, attending appointments and recommending further medical actions based on patient interactions. The Doctor 210 also reviews health reports and patient adherence to prescribed treatments.

[0081] The Appointment entity 212 includes attributes such as AppointmentID, PatientID, DoctorlD, AppointmentDate, and AppointmentTime. It includes operations such as bookAppointment(), confirmAppointment(), and sendAppointmentDetails(). The Appointment entity 212 is closely linked to the Report entity 214, which generates detailed health reports based on patient data and doctor reviews.

[0082] The Report entity 214 includes attributes like ReportID, PatientID, DoctorlD, ReportDetails, and ReportDate. It generates and sends reports to both patients and healthcare providers, facilitating informed medical decision-making.

[0083] The LLM Model entity 216 represents the large language model used in the system. It includes attributes like ModellD and operations such as analyzeHealthScreening(), generatePilllnformation(), predictHealthRisks(), recommendHealthcareVisit(), and provideMedicationlnformation(). The LLMModel 216 interacts with the Patient profile 204 to generate actionable health insights and recommendations.

[0084] The CNNModel entity 218 is responsible for identifying and processing pill images. It includes attributes such as ModellD and operations like identifyPill() andprocessPilllmage(). The CNNModel 218 collects and analyzes visual data from the Patient profile 204, providing accurate medication identification and classification.

[0085] The BiometricData entity 220 captures real-time physiological data from the patient, including attributes like BiometricDatalD, PatientID, HeartRate, BloodPressure, GlucoseLevel, BloodOxygenLevel, CalorieCount, ECG, and FallDetected. It includes the operation sendBiometricData(), which ensures continuous health monitoring and proactive health management.

[0086] The Medication entity 222 stores information about prescribed medications, including attributes such as MedicationlD, Name, Dosage, Instructions, SideEffects, and PatientID. It includes operations like trackMedication(), sendDosageReminder(), and sendAdherenceAlert(). The Medication entity 222 ensures that patients adhere to their prescribed treatment plans and provides timely reminders for medication intake.

[0087] The Notification entity 224 handles communication with patients, including attributes such as Notification! D, PatientID, Message, and NotificationDate. It includes the operation sendNotification(), which delivers alerts and reminders to patients about their health management activities.

[0088] The PredictiveAlert entity 226 generates alerts based on predictive analytics, including attributes like PredictiveAlertID, PatientID, Message, and AlertDate. It includes the operation generatePredictiveAlerts(), which identifies potential health risks and recommends timely interventions.

[0089] FIG. 2B illustrates the Design Class Diagram for the "Sihha Al" system (an example title for the system), detailing the relationships and interactions between various classes within the system. The SihhaAISystem class 228 integrates functionalities from multiple components to provide comprehensive healthcare management.

[0090] The Patient class 230 includes attributes such as name, userid, password, email, and phoneNumber, with methods like login(), capturePilllmage(), provideBiometricDataO, requestHealthScreening(), and confirmAppointmentBooking(). This class interacts with the SihhaAISystem class 228 to utilize the system's healthcare management capabilities.

[0091] The Doctor class 232 includes attributes such as name, userid, specialty, email, and phoneNumber, with methods like reviewHealthReport(), reviewPatientAdherence(), and prescribeMedication(). The Doctor class 232 uses the SihhaAISystem class 228 to access patient data, review reports, and manage treatment plans.

[0092] The SystemAdministrator class 234 includes attributes like adminld, name, email, and phoneNumber, with methods such as managellserAccess() and performSystemMaintenance(). The SystemAdministrator class 234 manages the overall system and ensures secure and efficient operation.

[0093] The SihhaAISystem class 228 includes methods like authenticatellser(), integrateBiometricDataO, analyzeData(), generateReport(), scheduleAppointment(), sendNotification(), trackMedication(), processPilllmage(), provideMedicationlnformation(), sendAdherenceAlerts(), and providePredictiveAnalytics(). This central class orchestrates the interactions between different components and ensures that the system functions cohesively.

[0094] The LLM Model class 236 includes methods such as analyzeHealthScreening(), generatePilllnformation(), predictHealthRisks(), recommendHealthcareVisit(), and provideMedicationlnformation(). This class interacts with the SihhaAISystem class 228 to generate actionable insights and recommendations based on patient data.

[0095] The CNNModel class 238 includes methods like identifyPill() and processPilllmage(). It interacts with the SihhaAISystem class 228 to provide accurate medication identification and classification based on pill images captured by patients.

[0096] The Database class 240 includes methods such as storeData(), retrieveData(), updateData(), and deleteData(). This class manages the storage and retrieval of user health data, ensuring data integrity and security.

[0097] The Appointment class 242 includes attributes such as appointmentld, appointmentDate, appointmentTime, patientld, and doctorld, with methods like bookAppointment(), confirmAppointment(), and sendAppointmentDetails(). It interacts with the SihhaAISystem class 228 to manage patient appointments and consultations.

[0098] The Medication class 244 includes attributes like medicationld, name, dosage, instructions, and sideEffects, with methods such as trackMedication(), sendDosageReminder(), and sendAdherenceAlert(). This class interacts with the SihhaAISystem class 228 to ensure patients adhere to their prescribed treatment plans.

[0099] The Report class 246 includes attributes such as reportld, patientld, doctorld, reportDetails, and reportDate, with methods like generateReport() and sendReport(). It interacts with the SihhaAISystem class 228 to provide detailed health reports to patients and healthcare providers.

[0100] The Notification class 248 includes attributes like notificationld, userid, message, and notificationDate, with methods such as sendNotification(). This class manages communication with patients and delivers timely health management alerts and reminders.

[0101] The HealthData class 250 includes attributes such as heartRate, bloodPressure, glucoseLevel, bloodOxygenLevel, calorieCount, ECG, and fallDetected, with methods like sendBiometricData(). This class captures and transmits real-time physiological data from patients, ensuring continuous health monitoring.

[0102] The PredictiveAnalytics class 252 includes methods such as analyzeHeartRate(), analyzeBloodPressure(), analyzeGlucoseLevel(), analyzeBloodOxygenLevel(), analyzeCalorieCount(), analyzeECG(), detectFall(), generatePredictiveAlerts(), and recommendHealthcareVisit(). This class interacts with the SihhaAISystem class 228 to provide predictive analytics and early warnings for potential health issues.

[0103] FIG. 3 illustrates a flow diagram of a generalized process in which a patient accesses a service provided by the system, highlighting the interaction between the CNN and LLM based on the patient's biometric and / or medication data.

[0104] The process begins with the patient accessing the system through the dedicated app on their user device (Step 300). The patient logs into the app using their credentials, which are authenticated by the system (Step 302). Once logged in, thepatient has various options, including the option to either capture an image of a medication or input their biometric data collected from wearable devices (Step 304).

[0105] If the patient chooses to capture an image of a medication, they use the app's interface to take a picture of the pill (Step 306). The captured image is then sent to the visual data processing module, where the CNN processes the image (Step 308). The CNN identifies the medication by analyzing features such as shape, color, and markings, and then classifies the pill (Step 310).

[0106] Concurrently, if the patient inputs biometric data, such as heart rate, blood pressure, or glucose levels, this data is collected and sent to the biometric data collection module (Step 312). The system continuously monitors and analyzes this biometric data to detect any abnormalities or potential health risks (Step 314). This set of steps is repeated continuously as long as the patient continues to upload / input their biometric data.

[0107] The outputs from both the CNN (medication identification) and the biometric data collection module (biometric analysis) are then forwarded to the natural language processing module, where the LLM processes this combined data (Step 316). The LLM generates a comprehensive analysis, including detailed information about the identified medication, its usage instructions, potential side effects, and any drug interactions based on the patient's current biometric data (Step 318).

[0108] Next, the LLM provides personalized medical advice and recommendations, which may include dosage reminders, adherence alerts, and health tips tailored to the patient's current health status and medication regimen (Step 320). This information is displayed to the patient through the app's user interface (Step 322), ensuring the patient receives timely and relevant healthcare guidance.

[0109] In case the LLM detects any critical health issues or significant medication interactions, it generates alerts and recommendations for further medical consultation. These alerts are then sent to the patient's healthcare provider (Step 324), who can review the patient's data and provide additional medical advice or adjustments to the treatment plan (Step 326).

[0110] The system logs all interactions and data into the secure databases for future reference and continuous learning. The continuous learning aspect allows the system to adapt and improve over time based on user interactions and feedback, enhancing the accuracy and relevance of its analyses and recommendations.

[0111] FIG. 4 illustrates a sequence diagram for providing a workflow involving the System to ensure patients adhere to their medication schedules. This workflow incorporates various applications of the system, such as pill identification, medication information provision, and adherence tracking.

[0112] The process begins with either the patient or doctor adding a medication schedule to the System (Step 400). This schedule outlines the medications the patient needs to take, including dosages and timings. The System then stores the medication schedule in its database (Step 402). This step ensures that the system has a record of what medications the patient needs to take and when. The system confirms that the medication schedule has been successfully stored (Step 404), ensuring data integrity and accuracy.

[0113] When it is time for the patient to take a medication, the patient captures an image of the pill using their device (Step 406). This image is sent to the System for identification. The System uses the CNN model to process the pill image and identify the medication (Step 408). This step ensures the patient is taking the correct pill as per the schedule. The system returns the details of the identified pill to the patient (Step 410). This information includes the name, dosage, and any other relevant details about the medication.

[0114] The pill information is displayed to the patient, confirming that they have the correct medication before ingestion (Step 412). If the patient requires additional information about the medication, they can request it through the app (Step 414). The System, utilizing the LLM model, provides comprehensive information about the medication (Step 416). This may include usage instructions, potential side effects, and other relevant data. The system then returns the requested medication information to the patient (Step 418), which is displayed to the patient to help them understand the medication better (Step 420).

[0115] At the scheduled times, the System sends a dosage reminder to the patient (Step 422). This reminder helps ensure that the patient adheres to the prescribed medication regimen. The patient confirms that they have taken the medication (Step 424), and this confirmation is sent back to the System. The system updates the adherence record to reflect that the patient has taken their medication as prescribed (Step 426) and confirms that the adherence record has been successfully updated (Step 428).

[0116] If the patient does not confirm medication intake, the System sends a missed dose alert (Step 430). This alert reminds the patient to take their medication or provides instructions on what to do if a dose is missed. The patient receives the missed dose alert (Step 432), prompting them to take the necessary action.

[0117] Periodically, the doctor reviews the patient's adherence records (Step 434). This review helps ensure that the patient is following the prescribed medication regimen. The System generates an adherence report based on the recorded data (Step 436). This report provides detailed insights into the patient's medication adherence over time. The system provides the adherence report to both the doctor and the patient (Step 438), helping both parties understand the patient's adherence patterns and make informed decisions about their treatment. Finally, the adherence report is displayed to the patient and doctor (Step 440), concluding the workflow.

[0118] FIGs 5A to 5D show various example sequences of steps for more detailed versions of the operations covered in the overall process of FIG.4.

[0119] FIG. 5A illustrates a sequence diagram for detailing the interaction between a patient, the System, a Convolutional Neural Network (CNN) model, and a Large Language Model (LLM) to achieve the task of identifying a pill and providing relevant information about it.

[0120] The process begins with the patient initiating a request for pill identification through the System (Step 500). This request can be made via a mobile application or a web interface. In response to the patient's request, the System prompts the patient to capture an image of the pill (Step 502). This step ensures that the system receives the necessary visual input for identification.

[0121] The patient then uses their device to take a picture of the pill and submits this image to the System (Step 504). The captured image serves as the primary data for the subsequent identification process. The System forwards the captured pill image to the CNN model (Step 506). The CNN model specializes in processing and analyzing visual data, which is crucial for accurately identifying the pill based on its physical characteristics.

[0122] The CNN model processes the pill image to identify distinguishing features such as shape, color, and markings (Step 508). This step involves complex image recognition algorithms inherent to CNNs. After processing the image, the CNN model returns the identification results to the System (Step 510). These results typically include the pill's name or identifying code.

[0123] Next, the System sends a request for detailed information about the identified pill to the LLM model (Step 512). This step leverages the LLM's capabilities in natural language processing and data retrieval. The LLM model generates comprehensive information about the pill, including its uses, dosage, potential side effects, and any other relevant details (Step 514). This information is curated from vast medical databases and literature.

[0124] The generated information is sent back from the LLM model to the System, completing the data retrieval process (Step 516). Finally, the System displays the pill identification and detailed information to the patient (Step 518). This allows the patient to understand what the pill is and how it should be used, ensuring proper medication management.

[0125] FIG. 5B illustrates a sequence diagram for outlining the interaction between a patient, a doctor, a hospital system, the System, and a Large Language Model (LLM) to conduct a health screening and manage appointments if necessary.

[0126] The process starts with the patient requesting a health screening through the System (Step 600). This request initiates the health screening process. The System begins the health screening by asking the patient a series of screening questions (Step 602). These questions are designed to gather relevant health information from the patient.

[0127] The system presents the screening questions to the patient (Step 604), who then provides answers (Step 606). These answers are critical inputs for the subsequent analysis. The System, utilizing the LLM model, analyzes the patient's responses to the screening questions (Step 608). This analysis helps in generating a comprehensive health report.

[0128] Based on the analysis of the patient's responses, the LLM model generates a detailed health report (Step 610). This report includes the patient's health status and any identified health risks or concerns. The generated health report is displayed to the patient (Step 612), providing them with an overview of their health based on the screening.

[0129] If the health screening indicates that an appointment with a healthcare provider is necessary, the system will recommend booking an appointment (Step 614). The patient confirms the need to book an appointment based on the recommendation (Step 616). The System interacts with the hospital system to book an appointment for the patient (Step 618). This involves selecting an appropriate time and healthcare provider based on the health report and the patient's needs.

[0130] The hospital system confirms the appointment booking and provides the necessary details to the System (Step 620). The System sends the confirmed appointment details back to the patient (Step 622), ensuring they have all the information needed for their upcoming appointment. The health report is also shared with the relevant healthcare provider or hospital system (Step 624). This ensures that the healthcare provider has access to the patient's health information prior to the appointment.

[0131] The patient, doctor, and hospital system review the health report to prepare for the upcoming appointment (Step 626). This step ensures that all parties are informed about the patient's health status. The healthcare provider reviews the health report to understand the patient's condition and prepare for the consultation (Step 628). The healthcare provider or hospital system provides the patient with access to the health report and any additional information or instructions relevant to their care (Step 630).

[0132] FIG. 50 illustrates a sequence diagram for outlining the process by which the System monitors a patient's biometric data, medication adherence, and daily calorie count to generate and send relevant alerts and notifications. This involves interaction between the patient, the System, and the integrated LLM (Large Language Model) and CNN (Convolutional Neural Network) models.

[0133] The process begins with the System continuously monitoring the patient's biometric data (Step 700). This can include vital signs such as heart rate, blood pressure, glucose levels, etc., collected from wearable devices or other health monitoring tools. The system uses the LLM and CNN models to analyze the monitored biometric data, checking for any irregularities or deviations from the patient's normal health parameters (Step 702).

[0134] If any irregularities are detected, the system identifies and flags these issues for further action (Step 704). This step ensures timely detection of potential health problems. Upon detecting any irregularities, the System sends an alert notification to the patient (Step 706). This alert informs the patient about the detected issue, prompting them to take necessary action or consult with a healthcare provider.

[0135] The System also tracks the patient's medication adherence (Step 708). This involves monitoring whether the patient is taking their medications as prescribed. If the system detects that the patient is not adhering to their medication schedule, it generates adherence alerts (Step 710). These alerts remind the patient to take their medications on time, ensuring better management of their health condition. The System sends these adherence alerts to the patient (Step 712), helping them stay on track with their prescribed medication regimen.

[0136] In addition to biometric data and medication adherence, the System monitors the patient's daily calorie intake (Step 714). This is important for patients managing conditions such as diabetes, obesity, or cardiovascular diseases. The system analyzes the patient's daily calorie count using the LLM and CNN models and generates notifications based on the analysis (Step 716). These notifications can help the patient maintain a balanced diet and adhere to their nutritional goals. The System sends daily calorie count notifications to the patient (Step 718). These notifications provide insights into their calorie intake, helping them make informed dietary choices.

[0137] FIG. 5D illustrates a sequence diagram for demonstrating how the System uses biometric data from a patient to generate predictive health alerts and recommend visits to healthcare professionals if necessary. This process involves interaction between the patient, the System, and the combined LLM (Large Language Model) and CNN (Convolutional Neural Network) models.

[0138] The process begins with the patient providing their biometric data to the System (Step 800). This data can include various health metrics such as heart rate, blood pressure, glucose levels, etc., typically collected through wearable devices or other health monitoring tools. The System forwards the biometric data to the integrated LLM and CNN models. These models collaboratively analyze the biometric data to identify patterns, trends, and potential health issues (Step 802).

[0139] Based on the analysis of the biometric data, the LLM and CNN models generate health predictions (Step 804). These predictions might include potential risks, early warnings of medical conditions, or trends indicating the progression of an existing condition. The System then displays these predictive alerts to the patient (Step 806). The alerts inform the patient about the identified risks or health trends, enabling them to take proactive steps in managing their health.

[0140] If the predictive analysis indicates a significant health concern that requires professional medical attention, the System recommends that the patient visit a healthcare professional (Step 808). This recommendation is based on the severity and nature of the predictions generated by the LLM and CNN models.

[0141] In some embodiments, the system includes a functionality that allows a parent to add dependents under their healthcare profile. This feature enables parents to manage and monitor the health data of their children through the system. A parent can create and link dependent profiles, allowing access to their children’s health records, medication schedules, biometric data, and interaction history. The user interface is configured to provide parents with the ability to view and control healthcare interactions for each dependent, ensuring comprehensive oversight and management. This functionality ensures that family health management is centralized, facilitating easier tracking of health metrics, appointment scheduling, and medication adherencefor children, while also providing necessary alerts and notifications to the parent regarding their dependents' health status.CONTROLLER / PROCESSOR COMPONENTS

[0142] A computing device or server as described herein can be any suitable type of computer. A computer may be a uniprocessor or multiprocessor machine. Accordingly, a computer may include one or more processors and, thus, the aforementioned computer system may also include one or more processors. Examples of processors include sequential state machines, microprocessors, microcontrollers, graphics processing units (GPUs), central processing units (CPUs), application processors, digital signal processors (DSPs), reduced instruction set computing (RISC) processors, systems on a chip (SoC), baseband processors, field programmable gate arrays (FPGAs), programmable logic devices (PLDs), gated logic, programmable control boards (PCBs), and other suitable hardware configured to perform the various functionality described throughout this disclosure.

[0143] Additionally, the computer may include one or more memories. Accordingly, the aforementioned computer systems may include one or more memories. A memory may include a memory storage device or an addressable storage medium which may include, by way of example, random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), electronically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), hard disks, floppy disks, laser disk players, digital video disks, compact disks, video tapes, audio tapes, magnetic recording tracks, magnetic tunnel junction (MTJ) memory, optical memory storage, quantum mechanical storage, electronic networks, and / or other devices or technologies used to store electronic content such as programs and data. In particular, the one or more memories may store computer executable instructions that, when executed by the one or more processors, cause the one or more processors to implement the procedures and techniques described herein. The one or more processors may be operably associated with the one or more memories so that the computer executable instructions can be provided to the one or more processors for execution. For example, the one or more processors may be operably associated to the one or more memories through one or more buses. Furthermore, the computermay possess or may be operably associated with input devices (e.g., a keyboard, a keypad, controller, a mouse, a microphone, a touch screen, a sensor) and output devices such as (e.g., a computer screen, printer, or a speaker).

[0144] The computer may advantageously be equipped with a network communication device such as a network interface card, a modem, or other network connection device suitable for connecting to one or more networks.

[0145] A computer may advantageously contain control logic, or program logic, or other substrate configuration representing data and instructions, which cause the computer to operate in a specific and predefined manner as, described herein. In particular, the computer programs, when executed, enable a control processor to perform and / or cause the performance of features of the present disclosure. The control logic may advantageously be implemented as one or more modules. The modules may advantageously be configured to reside on the computer memory and execute on the one or more processors. The modules include, but are not limited to, software or hardware components that perform certain tasks. Thus, a module may include, by way of example, components, such as, software components, processes, functions, subroutines, procedures, attributes, class components, task components, object-oriented software components, segments of program code, drivers, firmware, micro code, circuitry, data, and / or the like.

[0146] The control logic conventionally includes the manipulation of digital bits by the processor and the maintenance of these bits within memory storage devices resident in one or more of the memory storage devices. Such memory storage devices may impose a physical organization upon the collection of stored data bits, which are generally stored by specific electrical or magnetic storage cells.

[0147] The control logic generally performs a sequence of computer-executed steps. These steps generally require manipulations of physical quantities. Usually, although not necessarily, these quantities take the form of electrical, magnetic, or optical signals capable of being stored, transferred, combined, compared, or otherwise manipulated. It is conventional for those skilled in the art to refer to these signals as bits, values, elements, symbols, characters, text, terms, numbers, files, or the like. It should be kept in mind, however, that these and some other terms should be associated withappropriate physical quantities for computer operations, and that these terms are merely conventional labels applied to physical quantities that exist within and during operation of the computer based on designed relationships between these physical quantities and the symbolic values they represent.

[0148] It should be understood that manipulations within the computer are often referred to in terms of adding, comparing, moving, searching, or the like, which are often associated with manual operations performed by a human operator. It is to be understood that no involvement of the human operator may be necessary, or even desirable. The operations described herein are machine operations performed in conjunction with the human operator or user that interacts with the computer or computers.

[0149] It should also be understood that the programs, modules, processes, methods, and the like, described herein are but an exemplary implementation and are not related, or limited, to any particular computer, apparatus, or computer language. Rather, various types of general-purpose computing machines or devices may be used with programs constructed in accordance with some of the teachings described herein. In some embodiments, very specific computing machines, with specific functionality, may be required.CONCLUSION

[0150] Unless otherwise defined, all terms (including technical terms) used herein have the same meaning as commonly understood by one having ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and thepresent disclosure and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

[0151] The disclosed embodiments are illustrative, not restrictive. While specific configurations of the system for real-time healthcare management of the invention have been described in a specific manner referring to the illustrated embodiments, it is understood that the present invention can be applied to a wide variety of solutions which fit within the scope and spirit of the claims. There are many alternative ways of implementing the invention.

[0152] It is to be understood that the embodiments of the invention herein described are merely illustrative of the application of the principles of the invention. Reference herein to details of the illustrated embodiments is not intended to limit the scope of the claims, which themselves recite those features regarded as essential to the invention.

Claims

1. A system for real-time healthcare management, comprising: a) a data processing unit configured to receive and process multimodal data inputs, the data processing unit including: i) a visual data processing module comprising a Vision Transformer (ViT) architecture configured to receive, process, and classify images of medications by modeling spatial relationships and long-range dependencies, and output interpretable attention maps; ii) a natural language processing module comprising a transformer-based large language model (LLM) configured as a Large Action Model (LAM) using an agentic orchestration framework to receive, process, and generate natural language responses and dynamically interact with external tools including the ViT module, biometric analytics engine, and predictive models; b) a biometric data collection module configured to receive real-time biometric data from wearable or embedded devices; c) an integration module configured to: i) facilitate closed-loop communication between the ViT and the LLM such that outputs from the ViT are interpreted by the LLM and the LLM’s semantic context influences the ViT’s classification thresholds or attention mechanisms; ii) facilitate communication between the data processing unit and external healthcare systems through APIs; d) a user interface configured to: i) receive user inputs including pill images and natural language queries; ii) display processed data, system feedback, health alerts, and generated responses; e) a data storage unit configured to store multimodal inputs, processed outputs, and user-specific health data; f) a communication module configured to transmit and receive data between the system and external devices or cloud environments.

2. The system of claim 1 , wherein the visual data processing module comprises few-shot learning or transfer learning capabilities for rare medication classes.

3. The system of claim 1 , wherein the visual data processing module includes image preprocessing to resize, normalize, and augment pill images.

4. The system of claim 1 , wherein the natural language processing module utilizes multi-head attention, positional encodings, and domain-specific fine-tuning.

5. The system of claim 1 , wherein the biometric module integrates with wearable devices including smartwatches, fitness trackers, medical-grade wearables, and implants.

6. The system of claim 1 , wherein the integration module includes an API gateway and middleware to structure ViT outputs for the LLM.

7. The system of claim 1 , wherein the user interface includes responsive web / mobile interfaces with visualization and notifications.

8. The system of claim 1 , wherein the data storage unit includes encrypted databases and regulatory audit logging.

9. The system of claim 1 , wherein the ViT performs counterfeit detection, medication inventory management, and pill documentation.

10. The system of claim 1 , wherein the LLM manages scheduling, generates health reports, and tracks medication adherence.

11. The system of claim 1 , further comprising predictive analytics using RNNs for time-series health forecasting.

12. The system of claim 1 , wherein users provide feedback improving system performance.

13. The system of claim 1 , wherein LLM feedback modifies ViT thresholds and attention weighting.

14. The system of claim 1 , wherein biometric data undergoes calibration, fusion, and anomaly detection.

15. The system of claim 1 , wherein the integration module includes authentication, authorization, and data normalization.

16. The system of claim 1 , wherein the user interface includes symptom body maps and visualization tools.

17. The system of claim 1 , wherein the data storage unit supports archival, backup, and disaster recovery.

18. The system of claim 1 , wherein the ViT cross-references pill images with prescriptions to prevent medication errors.

19. The system of claim 1 , further comprising a patient engagement module delivering personalized health advice and motivational content.

20. The system of claim 1 , wherein the biometric module synchronizes with external EHR systems to maintain updated biometric records.

21. The system of claim 1 , wherein the ViT generates class activation maps and token-level attention outputs.

22. The system of claim 1 , wherein the LLM utilizes an agentic orchestration framework to invoke external tools based on task context.

23. The system of claim 1 , wherein semantic cues from LLM processing reduce false positives in ViT outputs.

24. The system of claim 1 , wherein the predictive engine forecasts health events over short, medium, and long-term intervals.

25. The system of claim 1 , wherein federated learning updates models without centralizing user data.

26. The system of claim 1 , wherein the user interface visualizes ViT attention maps for explainability.

27. The system of claim 1 , wherein the ViT processes both static images and live video frames.

28. The system of claim 1 , wherein the LLM identifies adverse medication reactions using cross-modal data.

29. The system of claim 1 , wherein system outputs are stored on a blockchain-based ledger.

30. The system of claim 1 , wherein the LLM is fine-tuned using reinforcement learning from human feedback.

31. The system of claim 1 , wherein confidence scores accompany system-generated alerts.

32. The system of claim 1 , wherein user-specific profiles influence LLM prompt templates.

33. The system of claim 1 , wherein low-confidence ViT classifications trigger human verification.

34. The system of claim 1 , wherein multimodal embeddings combine ViT, LLM, and biometric data for risk scoring.

35. The system of claim 1 , wherein system components support edge inference for offline operation.

36. The system of claim 1 , wherein the LAM module governs logic flow across system modules and APIs.

37. The system of claim 1 , wherein biometric time patterns correlate with medication adherence detected from images.

38. The system of claim 1 , wherein inference logs are visualized for healthcare provider review.

39. The system of claim 1 , wherein ViT patch embeddings are adjusted based on user-specific parameters.

40. The system of claim 1 , wherein system architecture follows microservices with CI / CD and regulatory compliance hooks.Statement under Article 19(1)The amended the claims to provide greater clarity and to more precisely define the technical contributions of the invention. These amendments refine the structure and operation of the system’s visual, language, biometric, and integration modules and more explicitly describe the interaction between these components. Additional dependent claims have been introduced to more fully reflect features disclosed in the original application, including the use of a Vision Transformer-based visual processing module, explainability mechanisms, agentic orchestration by the large language model, multimodal embedding techniques, predictive analytics, and data governance elements.The amended and added claims more accurately reflect the scope of the invention as originally disclosed and highlight the specific technical improvements achieved through the combination of multimodal Al processing, cross-module feedback, and enhanced data management architectures. These clarifications are intended to assist in the understanding of the invention and its contributions.

Images