Mobile terminal senile multi-course self-management and doctor-patient interaction data acquisition system

Abstract

The invention relates to the technical field of digital medical treatment, in particular to a mobile terminal senile multi-course autonomous management and doctor-patient interaction data acquisition system. The system comprises a personal digital twinning module used for constructing and optimizing a calculation model capable of simulating a multi-course interaction rule based on user multi-source health data; the situation awareness and embedded intervention module is used for sensing a user situation in real time according to a risk track deduced by the model, executing seamless fusion guiding intervention and generating first-class interaction data at the same time; and the doctor-patient collaborative decision-making module is used for providing a virtual adjustment tool based on digital twinning for the doctor and compiling the decision-making scheme into executable second-class interactive data. According to the invention, the problems of data islanding, stiff intervention, low doctor-patient cooperation efficiency and the like in multi-course management of old people are solved, and active, personalized and closed-loop intelligent health management is realized.

Description

Technical Field

[0001] This invention relates to the field of digital medical technology, and more specifically, to a mobile-based system for self-management of multiple disease courses in the elderly and for data collection of doctor-patient interactions. Background Technology

[0002] With the increasing aging of the population, the elderly often suffer from multiple chronic diseases coexisting, such as hypertension, diabetes, and coronary heart disease (multi-disease coexistence). Effective multi-disease co-management is crucial for controlling disease progression, improving quality of life, and reducing the burden of medical care. The development of mobile health technology has made home-based self-management possible.

[0003] However, existing mobile health management systems or applications have significant technical deficiencies in supporting multi-disease management in the elderly. Therefore, a mobile-based self-management and doctor-patient interaction data collection system for multi-disease management in the elderly is designed.

[0004] The existing technology has the following shortcomings, specifically: 1. Existing systems are mostly designed for single diseases, which cannot unify and comprehensively analyze multi-source heterogeneous data such as blood pressure, blood sugar, medication, and symptoms. This makes it difficult to reveal the interaction mechanisms between various chronic diseases, resulting in one-sided risk assessment and isolated or even conflicting management strategies.

[0005] 2. Data collection relies on tedious manual entry, and the interface design does not fully consider the abilities of the elderly, resulting in a high barrier to entry. Health reminders are mostly in the form of simple, passive, and easily ignored pop-ups and alarms, lacking intelligent guidance that integrates with the user's real-time context, leading to a poor user experience and low compliance.

[0006] 3. Doctor-patient communication is mostly limited to unstructured text and image consultations or simple Q&A. Doctors cannot easily obtain continuous, reliable, and structured data on patients' daily life and treatment, making it difficult to form a quantitative and continuous optimization management loop of "assessment-intervention-feedback-adjustment". The effect of remote management is limited. Summary of the Invention

[0007] The purpose of this invention is to provide a mobile-based system for self-management of multiple disease courses in the elderly and for data collection of doctor-patient interactions, in order to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention aims to provide a mobile terminal-based self-management and doctor-patient interaction data collection system for elderly patients with multiple diseases, including: a personal digital twin module, used to construct and continuously optimize a computational model that can dynamically simulate the pathophysiological interactions and evolutionary patterns among multiple chronic diseases based on the individualized multi-source health data of the target user.

[0009] The context-aware and embedded intervention module is connected to the personal digital twin module. It is used to perceive the digital and physical environment in which the user is located in real time, and generate and inject seamlessly integrated guided intervention operations into the environment based on the health risk trajectory deduced by the calculation model. At the same time, it generates the first type of interactive data containing the operation context and user response.

[0010] The doctor-patient collaborative decision-making module is connected to the personal digital twin module. It is used to receive and visualize the first type of interactive data, provide doctors with virtual adjustment and simulation tools based on the computing model, and compile the generated decision scheme into the second type of interactive data that can be parsed and executed by the user terminal.

[0011] The second type of interactive data is used to drive the context-aware and embedded intervention module to adjust its intervention logic, while the first type of interactive data is fed back to the personal digital twin module for model calibration.

[0012] As a further improvement to this technical solution, the multi-source health data includes at least vital sign monitoring data, medication and treatment record data, subjective symptom and behavior log data, and environmental and contextual data.

[0013] As a further improvement to this technical solution, the personal digital twin module includes a general pathophysiological model unit and a federated learning personalization unit.

[0014] Among them, the general pathophysiological model unit is pre-built with an initial computational framework based on medical knowledge graphs that reflects the mutual influence of various chronic diseases in the elderly.

[0015] The federated learning personalization unit is used to optimize and train the parameters of the initial computing framework locally on the user terminal using the user's private data and the first type of interactive data, and to update global knowledge through encrypted gradient uploading and aggregation, so as to realize the personalized evolution of the model and keep the original data in the domain.

[0016] As a further improvement to this technical solution, the context awareness and embedded intervention module includes a multi-source context awareness unit, a non-sensory intervention strategy execution unit, and a multimodal compliance verification unit.

[0017] Among them, the multi-source context-aware unit is used to acquire real-time user behavior, application usage, and physical environment context.

[0018] The non-intrusive intervention strategy execution unit has a pre-set micro-strategy library associated with specific risks and situations. When the triggering conditions are met, it executes actions to guide behavior by adjusting information presentation or environmental state.

[0019] The multimodal compliance verification unit is used to generate a quantitative compliance confidence score for the user's claimed completion of operation-related health management behaviors through multi-source signal acquisition and cross-analysis.

[0020] As a further improvement to this technical solution, the first type of interactive data is specifically implemented as follows: Step 1: Context capture and task triggering: When the non-intrusive intervention strategy execution unit matches the real-time context with the computational model and executes a guided operation, it synchronously records the metadata of the operation and the current environmental context to form an initial log.

[0021] Step 2: Multimodal compliance verification: For subsequent user statements or behaviors associated with the operation, the multimodal compliance verification unit is invoked to perform multi-source signal acquisition and cross-analysis to generate a quantified compliance confidence score.

[0022] Step 3: Data Association and Encapsulation: The initial logs, the compliance confidence scores, and any proactive feedback information submitted by the user through the preset interface are associated according to a unified time sequence and event ID, and encapsulated into a structured data packet.

[0023] Step 4: Local storage and synchronization: The structured data packets are stored locally as traceable event records, and synchronized to the doctor-patient collaborative decision-making module as the first type of interactive data according to preset rules or triggering conditions.

[0024] As a further improvement to this technical solution, the doctor-patient collaborative decision-making module includes a digital sandbox simulation unit and a digital prescription compilation unit.

[0025] The digital sand table simulation unit is used to receive treatment adjustment hypotheses input by doctors, extrapolate them on a copy of the computational model, and demonstrate the simulated impact on future multi-stage disease indicators.

[0026] The digital prescription compilation unit is used to compile the optimized plan confirmed by the doctor into a set of machine-executable instructions containing specific tasks, triggering conditions and verification rules, forming the second type of interactive data.

[0027] As a further improvement to this technical solution, the second type of interactive data is specifically implemented as follows: Step 1: Decision Input and Model Loading: In response to the treatment adjustment assumptions input by the healthcare provider in the digital sandbox simulation unit, a copy of the current user's computational model is loaded into the personal digital twin module.

[0028] Step 2: Multi-indicator simulation: Execute the adjustment assumptions on the copy, calculate the dynamic impact on multiple key health indicators of the user in the future preset time period, and generate a simulation report.

[0029] Step 3: Solution Optimization and Confirmation: Based on the comparative analysis of simulation reports of multiple adjustment assumptions, the healthcare provider confirms or optimizes the final intervention plan.

[0030] Step 4: Instruction compilation and encapsulation: The confirmed final intervention plan is compiled into a machine-executable instruction set containing specific task items, triggering conditions, execution parameters and success criteria through the digital prescription compilation unit, and digitally signed and encapsulated to form the second type of interactive data packet.

[0031] Step 5: Secure distribution and synchronization: Securely distribute the second type of interactive data packet to the corresponding user's mobile terminal and synchronously update the intervention plan record in the doctor collaboration platform.

[0032] As a further improvement to this technical solution, it also includes: a report generation module, used to automatically generate a periodic multi-disease-course health management comprehensive report based on the output of the personal digital twin module, historical first-type interaction data and historical second-type interaction data.

[0033] The report should include at least: an analysis of the changing trends of key physiological indicators, an analysis of the implementation of intervention tasks, an evolution of risk assessment across the course of the disease, and a summary of the phased management effects.

[0034] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By constructing a personal digital twin computing model, the system dynamically simulates the pathophysiological interactions among multiple chronic diseases of users, enabling quantitative assessment and forward-looking projection of comprehensive risks across disease courses, and providing a core intelligent engine for formulating collaborative and precise management strategies.

[0035] 2. By leveraging context awareness and embedded intervention modules, health guidance is intelligently and seamlessly integrated into the user's current digital and physical environment (e.g., adjusting the application interface), greatly enhancing acceptance. Combined with multimodal compliance verification, the authenticity of behavioral data is ensured, significantly improving management compliance and data quality.

[0036] 3. By defining and circulating standardized first and second types of interactive data, and with the help of digital sand table simulation and digital prescription compilation tools, a complete, quantitative, and automated management closed loop of "patient data feedback → doctor simulation decision-making → automatic issuance and execution of instructions → effect feedback" has been realized, which greatly improves the accuracy and efficiency of remote collaboration. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Example: Please refer to Figure 1 As shown, a mobile-based self-management and doctor-patient interaction data collection system for elderly patients with multiple diseases is provided, including: a personal digital twin module, which is used to build and continuously optimize a computational model that can dynamically simulate the pathophysiological interactions and evolution of multiple chronic diseases based on the individualized multi-source health data of the target user.

[0041] This computational model is built and continuously optimized around the core principle of "medical knowledge-driven + data-adaptive iteration": In the construction phase, a general computational framework for the interaction of multiple chronic disease pathophysiology is first established based on a medical knowledge graph. Then, multi-source health data is input for preprocessing and parameter initialization, and an initial personalized model is formed through historical backtesting. In the optimization phase, a closed-loop mechanism is used. On the one hand, lightweight parameter fine-tuning (adapting to intervention responses, contextual influences, etc.) is performed locally on the user's terminal using the first type of interaction data. On the other hand, federated learning is used to achieve encrypted gradient uploading and global aggregation updates (ensuring data privacy and improving model generalization ability). Simultaneously, the model structure is periodically calibrated and adaptively adjusted to cope with changes in the user's health status. Ultimately, this achieves dynamic and accurate simulation of the interaction and evolution of multiple chronic disease pathophysiology, providing support for contextualized intervention and collaborative doctor-patient decision-making. For example, it can be implemented in the form of a discrete-time state-space model, updating the state once at each time step (e.g., 15 minutes).

[0042] In one specific embodiment, the multi-source health data includes at least vital sign monitoring data, medication and treatment record data, subjective symptom and behavior log data, and environmental and contextual data.

[0043] Vital signs monitoring data comes from periodic or continuous physiological parameter measurements from wearable devices or home medical devices.

[0044] Medication and treatment record data, including drug name, dosage, medication time, and adherence record.

[0045] Subjective symptom and behavior log data are collected through an age-friendly interactive interface, which includes user-reported symptoms, feelings, and daily life events.

[0046] Environmental and contextual data refers to physical environment and behavioral contextual information related to user health, obtained through mobile terminal sensors or IoT devices.

[0047] In one specific embodiment, the personal digital twin module includes a general pathophysiological model unit and a federated learning personalization unit.

[0048] Among them, the general pathophysiological model unit is pre-built with an initial computational framework based on medical knowledge graphs that reflects the mutual influence of various chronic diseases in the elderly.

[0049] The medical knowledge graph transforms the pathophysiological mechanisms of common chronic diseases in the elderly (such as hypertension, diabetes, coronary heart disease, and chronic kidney disease) into a "node-edge" structure. Disease types serve as nodes, and pathophysiological interactions (promotion / inhibition / synergy) serve as edges. The weight of each edge corresponds to an initial quantification of the interaction strength (based on evidence-based medicine).

[0050] The federated learning personalization unit is used to optimize and train the parameters of the initial computing framework locally on the user terminal using the user's private data and the first type of interactive data, and to update global knowledge through encrypted gradient uploading and aggregation, so as to realize the personalized evolution of the model and keep the original data in the domain.

[0051] The context-aware and embedded intervention module is connected to the personal digital twin module. It is used to perceive the digital and physical environment in which the user is located in real time, and generate and inject seamlessly integrated guided intervention operations into the environment based on the health risk trajectory deduced by the calculation model. At the same time, it generates the first type of interactive data containing the operation context and user response.

[0052] The health risk trajectory refers to an output of the personal digital twin computing model, namely a dynamic and continuous prediction of the user's health status (especially the risk of deterioration) over a future period. This can be a time series curve (such as a prediction of blood glucose changes and periods of low blood glucose risk in the next 24 hours) or a probabilistic event chain (such as "current blood pressure continues to rise, the probability of headache symptoms occurring in the next 48 hours is X%, which in turn leads to a decrease in sleep quality is Y%)). For example, the computing model, based on the user's current blood glucose level, insulin medication time, and recent dietary log, calls its internal endocrine metabolism sub-model to deduce the blood glucose change trajectory in the next 6 hours and identifies high-risk periods where blood glucose levels may be below 4.0 mmol / L. This period is a specific 'health risk trajectory'.

[0053] The "environment" here refers to the carrier of the intervention operation, encompassing the user's physical and digital environments. The digital environment refers to the state of the user's mobile terminal and its application layer. For example, the applications the user is currently actively using (such as WeChat, Douyin, and food delivery apps), the device time (morning / late night), network status, and screen on / off status.

[0054] Physical environment: refers to the real-world state surrounding the user. For example: location determined by GPS / Wi-Fi positioning (home, park, supermarket), environmental data obtained by IoT devices (bedroom light brightness, living room temperature), and state sensed by mobile phone sensors (a stationary mobile phone may indicate that the user is resting).

[0055] Seamless integration: This means that the intervention does not appear as an abrupt pop-up alert, but rather blends naturally with the user's current task or environment, reducing cognitive burden.

[0056] Guided intervention: This refers to interventions whose purpose is not to "command," but to guide users to make healthier choices spontaneously by providing convenience, adjusting options, or offering gentle reminders.

[0057] For example: "When the model anticipates a user's risk of prolonged sitting and detects that the user is continuously swiping in a short video application (digital environment), the system uses the accessibility service interface to overlay a semi-transparent 'Get Up and Move' card at the bottom of the application interface, and provides a button to play broadcast gymnastics music with one click. This operation does not interrupt the video stream, achieving 'seamless integration' and 'guidance'."

[0058] Operational context records background information related to the entire process of guided intervention operations. It is used to accurately describe the "ins and outs" of the intervention operation and provides traceable evidence for the generation of subsequent first-type interactive data, model calibration (feedback to the personal digital twin module), and collaborative decision-making between doctors and patients. It usually includes: Triggering context: Why was this intervention triggered? Such as the specific risk trajectory content and risk level triggered.

[0059] Environmental snapshot: The specific state of the "environment" at the time of the intervention (a snapshot of the above digital / physical environment is recorded).

[0060] Operation details: What specific intervention command did the system execute (e.g., inserting the ZZ element into the YY interface of the XX application).

[0061] Timestamp: Precise trigger and execution time.

[0062] In one specific embodiment, the context awareness and embedded intervention module includes a multi-source context awareness unit, a non-intrusive intervention strategy execution unit, and a multimodal compliance verification unit.

[0063] Among them, the multi-source context-aware unit is used to acquire real-time user behavior, application usage, and physical environment context.

[0064] The non-intrusive intervention strategy execution unit has a pre-set micro-strategy library associated with specific risks and situations. When the triggering conditions are met, it executes actions to guide behavior by adjusting information presentation or environmental state.

[0065] Construction of a micro-strategy library linking specific risks and situations: Based on medical evidence and the behavioral habits of elderly users, we first sort out the correlation mapping relationship between common health risk types of chronic diseases in the elderly (such as blood sugar fluctuations, sudden rise in blood pressure, missed medication, etc.) and corresponding typical situations (home / outdoor, before / after meals, rest / activity, etc.). Then, based on this mapping, we design targeted micro-strategies (including information presentation methods, environmental adjustment actions, guiding words, etc.) for each type of "risk-situation" combination. At the same time, we pre-set strategy execution parameters (such as prompt timing, presentation duration, device linkage rules). Finally, we continuously iterate and optimize the adaptability of the strategies by combining historical intervention data and user feedback to form a structured and dynamically updatable micro-strategy library.

[0066] The triggering conditions specifically refer to a set of multi-dimensional conditions preset in the micro-policy library that must be met simultaneously. These conditions primarily fall into three categories: first, health risk conditions derived from the personal digital twin module's computational model (such as matching specific risk types, risk levels reaching preset thresholds, and risks occurring within the expected time window); second, real-time contextual conditions acquired by the multi-source contextual awareness unit (such as the user's current physical environment, ongoing behavior, application usage status, and smart device online status, etc., matching these contextual conditions with the micro-policy); and third, basic execution conditions (such as normal user terminal devices, smooth data transmission, and no user-defined intervention blocking periods). The seamless intervention strategy is only triggered when all three conditions are met.

[0067] The "seamless intervention strategy" specifically refers to a set of strategies designed with low operational burden and no behavioral interference for elderly users as its core design principle. Based on the "risk-situation" correlation mapping relationship in the micro-strategy library, it implicitly guides users to complete health management behaviors in a lightweight manner that seamlessly adapts to the user's current digital / physical environment and real-time behavior when the trigger conditions are met. Its core feature is "seamless"—it does not interrupt the user's current activity (such as watching TV, chatting, or doing housework), does not require the user to learn complex operations, and conveys the guidance intention only through optimized information presentation (such as semi-transparent prompts and gentle voice) or minor adjustments to the environmental state (such as smart device linkage). At the same time, it takes into account the targeted and humanized nature of the intervention, avoids user resistance caused by forced instructions, and ultimately helps avoid potential hidden dangers in the health risk trajectory with extremely low user perception.

[0068] The multimodal compliance verification unit is used to generate a quantitative compliance confidence score for the user's claimed completion of operation-related health management behaviors through multi-source signal acquisition and cross-analysis.

[0069] Quantitative compliance confidence score generation: Multi-source signals (such as smart device monitoring data, user behavior logs, environmental sensing data, terminal operation records, etc.) related to the health management behaviors claimed by users are collected through a multimodal compliance verification unit. Cross-validation and consistency analysis are performed on the multi-source signals (such as comparing the smart pillbox opening record with the user's claimed medication behavior, and matching exercise device data with exercise behavior statements). Then, the confidence weight of each signal is combined (such as the weight of device monitoring data is higher than that of subjective records). A quantitative score in the range of 0-1 is calculated by a preset algorithm (such as weighted summation, Bayesian inference). The higher the score, the higher the confidence of the user's behavior compliance.

[0070] In one specific embodiment, the first type of interactive data is implemented as follows: Step 1: Context capture and task triggering: When the non-intrusive intervention strategy execution unit matches the real-time context with the computational model and executes a guided operation, it synchronously records the metadata of the operation and the current environmental context to form an initial log.

[0071] Step 2: Multimodal compliance verification: For subsequent user statements or behaviors associated with the operation, the multimodal compliance verification unit is invoked to perform multi-source signal acquisition and cross-analysis to generate a quantified compliance confidence score.

[0072] Step 3: Data Association and Encapsulation: The initial logs, the compliance confidence scores, and any proactive feedback information submitted by the user through the preset interface are associated according to a unified time sequence and event ID, and encapsulated into a structured data packet.

[0073] Step 4: Local storage and synchronization: The structured data packets are stored locally as traceable event records, and synchronized to the doctor-patient collaborative decision-making module as the first type of interactive data according to preset rules or triggering conditions.

[0074] The doctor-patient collaborative decision-making module is connected to the personal digital twin module. It is used to receive and visualize the first type of interactive data, provide doctors with virtual adjustment and simulation tools based on the computing model, and compile the generated decision scheme into the second type of interactive data that can be parsed and executed by the user terminal.

[0075] The second type of interactive data is used to drive the context-aware and embedded intervention module to adjust its intervention logic, while the first type of interactive data is fed back to the personal digital twin module for model calibration. This forms a closed-loop management system with the personal digital twin at its core and the two types of doctor-patient interactive data as the link.

[0076] In one specific embodiment, the doctor-patient collaborative decision-making module includes a digital sandbox simulation unit and a digital prescription compilation unit.

[0077] The digital sand table simulation unit is used to receive treatment adjustment hypotheses input by doctors, extrapolate them on a copy of the computational model, and demonstrate the simulated impact on future multi-stage disease indicators.

[0078] The digital prescription compilation unit is used to compile the optimized plan confirmed by the doctor into a set of machine-executable instructions containing specific tasks, triggering conditions and verification rules, forming the second type of interactive data.

[0079] In one specific embodiment, the second type of interactive data is implemented as follows: Step 1: Decision Input and Model Loading: In response to the treatment adjustment assumptions input by the healthcare provider in the digital sandbox simulation unit, a copy of the current user's computational model is loaded into the personal digital twin module.

[0080] Step 2: Multi-indicator simulation: Execute the adjustment assumptions on the copy, calculate the dynamic impact on multiple key health indicators of the user in the future preset time period, and generate a simulation report.

[0081] Step 3: Solution Optimization and Confirmation: Based on the comparative analysis of simulation reports of multiple adjustment assumptions, the healthcare provider confirms or optimizes the final intervention plan.

[0082] Step 4: Instruction compilation and encapsulation: The confirmed final intervention plan is compiled into a machine-executable instruction set containing specific task items, triggering conditions, execution parameters and success criteria through the digital prescription compilation unit, and digitally signed and encapsulated to form the second type of interactive data packet.

[0083] Step 5: Secure distribution and synchronization: Securely distribute the second type of interactive data packet to the corresponding user's mobile terminal and synchronously update the intervention plan record in the doctor collaboration platform.

[0084] In one specific embodiment, it further includes: a report generation module, used to automatically generate a periodic multi-disease-course health management comprehensive report based on the output of the personal digital twin module, historical first-type interaction data and historical second-type interaction data.

[0085] The report should include at least: an analysis of the changing trends of key physiological indicators, an analysis of the implementation of intervention tasks, an evolution of risk assessment across the course of the disease, and a summary of the phased management effects.

[0086] The generation of a comprehensive multi-disease-course health management report involves first performing time-series alignment, deduplication, cleaning, and structured integration of multiple sources of data, including the health risk trajectory and model simulation results output by the individual's digital twin module, the intervention operation context, user response records, and compliance confidence scores from historical first-type interaction data, and the doctor's decision-making plans and machine-executable instructions from historical second-type interaction data. Then, based on preset report dimensions (trends in key physiological indicators, implementation status of intervention tasks, evolution of cross-disease-course risk assessment, and phased management effects), professional interpretation is achieved through data mining analysis (such as trend fitting, compliance statistics, and risk association inference) combined with medical knowledge graphs. Finally, the report is automatically formatted in a concise chart, plain text, and highlighted format that is easy for elderly users to understand, generating periodic (e.g., weekly / monthly) comprehensive reports to provide accurate and intuitive reference for users' self-management of health and doctor-patient communication.

[0087] This example uses Mr. Zhang (65 years old), an elderly user with hypertension and type 2 diabetes, to illustrate in detail how the system works collaboratively throughout a complete management cycle.

[0088] 1. Initial Configuration and Data Acquisition With the assistance of his children, Mr. Zhang installed the client application of this system on his smartphone and completed the following configuration: Device connection: The system automatically paired and connected to Mr. Zhang's smart blood pressure monitor and blood glucose meter via Bluetooth.

[0089] Information entry: Using voice assistance and a simplified form, the current medication regimen was entered: taking one tablet of the antihypertensive drug "Amlodipine" every morning and injecting insulin before each of the three daily meals.

[0090] Authorization settings: Mr. Zhang authorizes the system to obtain his location information when necessary and to provide health guidance when using food delivery and shopping apps.

[0091] The system has begun continuously collecting Mr. Zhang's individualized multi-source health data: Vital signs monitoring data: Every morning, the smart blood pressure monitor automatically measures blood pressure (e.g., 145 / 92 mmHg) and synchronizes it to the mobile phone; after each of the three meals, Mr. Zhang uses a blood glucose meter to measure fingertip blood glucose and it is automatically synchronized (e.g., blood glucose 2 hours after breakfast is 10.5 mmol / L).

[0092] Behavioral log data: Mr. Zhang recorded his subjective feelings by saying "I feel a little dizzy today" via voice or by selecting "poor sleep quality" through the interface.

[0093] Environmental and contextual data: The mobile phone GPS shows that Mr. Zhang mainly stays in the community during the day and stays at home at night; the motion sensor records that his average daily steps are about 3,000.

[0094] 2. Construction and Risk Simulation of Personal Digital Twin Module The personal digital twin module begins operating based on the above data: Model initialization: The module calls the general pathophysiological model unit to load a pre-built initial computational framework that includes the interaction between hypertension and diabetes.

[0095] Personalized Modeling: The federated learning personalized unit uses Mr. Zhang's blood pressure, blood sugar, medication time, and subjective symptoms data from the past week to train and adjust the initial model parameters locally on his mobile phone, generating a preliminary, personalized "digital twin" model for Mr. Zhang. This model can simulate the association between two diseases. For example, the simulation shows that if blood pressure remains high, even small changes in renal perfusion may lead to a decrease in insulin sensitivity of about 5% within the next 3 days, thereby exacerbating blood sugar fluctuations.

[0096] Risk trajectory projection: One evening, based on Mr. Zhang's high post-lunch blood sugar, low daytime activity level, and complaint of dizziness, the model projected a risk trajectory of abnormally high nocturnal blood pressure within the next 12 hours, with a probability of 65%. This risk was sent to the context awareness and embedded intervention module in real time.

[0097] 3. Execution and Validation of Context-Aware and Embedded Interventions Upon receiving a risk warning, the context awareness and embedded intervention module initiates the following process: Context awareness: The multi-source context awareness unit detected that it was 8 p.m., and Mr. Zhang was sitting on the sofa in the living room with a short video application running on his mobile phone.

[0098] Strategy Matching and Execution: The seamless intervention strategy execution unit matches a strategy from the micro-strategy library based on the combination of "nighttime blood pressure risk" and "user in a home leisure setting." The strategy's action is to gently push a banner notification containing health advice at the bottom of the user's current application interface without interrupting video playback. The system then executes the strategy, and a prompt appears at the bottom of Mr. Zhang's phone screen: "We detected that you have been less active today. We suggest you get up and move around for 10 minutes now to help stabilize your nighttime blood pressure. Click to play the guided video for a relaxing walk in the living room." Compliance verification and data generation: Mr. Zhang saw the prompt, clicked to play the guided video, and followed it to complete a 10-minute indoor walk.

[0099] Step 1 (Context Capture): The system records the metadata (trigger risk ID, execution strategy ID) and context snapshot (time, location, front-end application) of this intervention.

[0100] Step 2 (Multimodal Validation): Regarding Mr. Zhang's statement that he "completed the activity," the multimodal compliance validation unit initiated validation: it analyzed the phone's accelerometer data during the activity to identify a regular walking pattern; simultaneously, it analyzed system logs to confirm that the guidance video was played in its entirety. Through fusion analysis, a compliance confidence score of 92 points (high score) was generated for this activity.

[0101] Steps 3 and 4 (data encapsulation and synchronization): The system encapsulates the metadata of this intervention, the context snapshot, the confidence score of 92 points, and the fact that Mr. Zhang did not provide any additional feedback into a structured Type I interactive data package, stores it locally, and synchronizes it to the doctor-patient collaborative decision-making module in the cloud.

[0102] 4. Collaborative decision-making between doctors and patients and digital prescription issuance A week later, Mr. Zhang's family doctor, Dr. Li, logged into the doctor-patient collaborative decision-making module via computer.

[0103] Data Review: Dr. Li saw a summary of Mr. Zhang's first type of interaction data over the past week on the interface, including blood pressure and blood sugar trend charts, multiple intervention records, and his high compliance score.

[0104] Simulated Decision-Making: Dr. Li noticed that Mr. Zhang's morning blood pressure remained high. In the digital sandbox simulation unit, he input a hypothesis: "Change the amlodipine dosage time from morning to bedtime." The system immediately loaded a copy of Mr. Zhang's personal digital twin model, simulating the impact of this change on blood pressure (especially morning peak blood pressure) and blood sugar over the next week. The simulation report showed that the adjusted morning peak blood pressure was expected to decrease by 8 mmHg, with no significant adverse effect on blood sugar.

[0105] Digital Prescription Generation and Issuance: Based on the simulation results, Dr. Li confirmed the adjustment plan. The digital prescription compilation unit compiled this plan into a Type II interactive data (structured digital prescription), which included: the new medication time (before bedtime), duration for one week, daily reminders, and a requirement to confirm via mobile phone after medication. The digital prescription was securely issued to Mr. Zhang's mobile phone.

[0106] 5. Closed-loop execution and report generation Prescription Execution: Mr. Zhang received a medication reminder on his phone before bed. After taking the medication, he confirmed it via voice. The system recorded this confirmation and initiated multimodal verification (such as analyzing the sound of the medicine bottle being opened). This new medication behavior data was fed back to the system as new type 1 interaction data.

[0107] Model calibration: Newly generated medication time data and subsequent blood pressure measurements are fed back to the personal digital twin module as the first type of interactive data to calibrate the model and make its future predictions more accurate.

[0108] Periodic report generation: At the end of the month, the report generation module runs automatically. It summarizes Mr. Zhang's digital twin outputs and all first and second category interaction data from the past month, generates a "Comprehensive Report on Multi-Disease Course Health Management," and sends it to Dr. Li and Mr. Zhang.

[0109] The report shows that Mr. Zhang's morning blood pressure began to decline in the second week after adjusting his medication schedule, and the rate of achieving the target increased to 70% this month.

[0110] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A mobile-based self-management and doctor-patient interaction data collection system for elderly patients with multiple disease courses, characterized in that: include: The personal digital twin module is used to build and continuously optimize a computational model that can dynamically simulate the pathophysiological interactions and evolution of various chronic diseases based on the individualized multi-source health data of the target user. The context awareness and embedded intervention module is connected to the personal digital twin module. It is used to perceive the digital and physical environment in which the user is located in real time, and generate and inject seamlessly integrated guided intervention operations into the environment based on the health risk trajectory deduced by the calculation model. At the same time, it generates the first type of interactive data containing the operation context and user response. The doctor-patient collaborative decision-making module is connected to the personal digital twin module and is used to receive and visualize the first type of interactive data, provide doctors with virtual adjustment and simulation tools based on the computing model, and compile the generated decision scheme into a second type of interactive data that can be parsed and executed by the user terminal. The second type of interactive data is used to drive the context-aware and embedded intervention module to adjust its intervention logic, while the first type of interactive data is fed back to the personal digital twin module for model calibration.

2. The mobile terminal elderly multi-disease self-management and doctor-patient interaction data collection system according to claim 1, characterized in that: The multi-source health data includes at least vital sign monitoring data, medication and treatment record data, subjective symptom and behavior log data, and environmental and contextual data.

3. The mobile terminal elderly multi-disease self-management and doctor-patient interaction data collection system according to claim 1, characterized in that: The personal digital twin module includes a general pathophysiological model unit and a federated learning personalization unit; Among them, the general pathophysiological model unit is pre-built with an initial computational framework based on medical knowledge graph that reflects the mutual influence of various chronic diseases in the elderly; The federated learning personalization unit is used to optimize and train the parameters of the initial computing framework locally on the user terminal using the user's private data and the first type of interactive data, and to update global knowledge through encrypted gradient uploading and aggregation, so as to realize the personalized evolution of the model and keep the original data in the domain.

4. The mobile terminal elderly multi-disease self-management and doctor-patient interaction data collection system according to claim 3, characterized in that: The context awareness and embedded intervention module includes a multi-source context awareness unit, a non-intrusive intervention strategy execution unit, and a multimodal compliance verification unit; Among them, the multi-source context-aware unit is used to acquire real-time user behavior, application usage and physical environment context; The non-intrusive intervention strategy execution unit is pre-loaded with a micro-strategy library associated with specific risks and situations. When the triggering conditions are met, it executes actions to guide behavior by adjusting information presentation or environmental state. The multimodal compliance verification unit is used to generate a quantitative compliance confidence score for the user's claimed completion of operation-related health management behaviors through multi-source signal acquisition and cross-analysis.

5. The mobile terminal-based elderly multi-disease self-management and doctor-patient interaction data collection system according to claim 1, characterized in that: The specific implementation method for the first type of interactive data is as follows: Step 1: Context capture and task triggering: When the non-intrusive intervention strategy execution unit matches the real-time context with the computational model and executes a guided operation, it synchronously records the metadata of the operation and the current environmental context to form an initial log; Step 2: Multimodal compliance verification: For subsequent user statements or behaviors associated with the operation, the multimodal compliance verification unit is invoked to perform multi-source signal acquisition and cross-analysis to generate a quantified compliance confidence score. Step 3: Data Association and Encapsulation: The initial log, the compliance confidence score, and any proactive feedback information submitted by the user through the preset interface are associated according to a unified time sequence and event ID, and encapsulated into a structured data packet; Step 4: Local storage and synchronization: The structured data packets are stored locally as traceable event records, and synchronized to the doctor-patient collaborative decision-making module as the first type of interactive data according to preset rules or triggering conditions.

6. The mobile terminal-based elderly multi-disease self-management and doctor-patient interaction data collection system according to claim 1, characterized in that: The doctor-patient collaborative decision-making module includes a digital sand table simulation unit and a digital prescription compilation unit; The digital sand table simulation unit is used to receive treatment adjustment hypotheses input by doctors, extrapolate them on a copy of the computational model, and demonstrate the simulated impact on future multi-stage disease indicators. The digital prescription compilation unit is used to compile the optimized plan confirmed by the doctor into a set of machine-executable instructions containing specific tasks, triggering conditions and verification rules, forming the second type of interactive data.

7. The mobile terminal elderly multi-disease self-management and doctor-patient interaction data collection system according to claim 6, characterized in that: The second type of interactive data is implemented as follows: Step 1: Decision Input and Model Loading: In response to the treatment adjustment assumptions input by the healthcare provider in the digital sandbox simulation unit, load a copy of the current user's computational model in the personal digital twin module; Step 2: Multi-indicator simulation: Execute the adjustment assumptions on the copy, calculate the dynamic impact on multiple key health indicators of the user in the future preset time period, and generate a simulation report; Step 3: Solution Optimization and Confirmation: Based on the comparative analysis of simulation reports of multiple adjustment assumptions, the healthcare provider confirms or optimizes the final intervention plan; Step 4: Instruction compilation and encapsulation: The confirmed final intervention plan is compiled into a machine-executable instruction set containing specific task items, triggering conditions, execution parameters and success criteria through the digital prescription compilation unit, and digitally signed and encapsulated to form the second type of interactive data packet; Step 5: Secure distribution and synchronization: Securely distribute the second type of interactive data packet to the corresponding user's mobile terminal and synchronously update the intervention plan record in the doctor collaboration platform.

8. The mobile terminal elderly multi-disease self-management and doctor-patient interaction data collection system according to claim 1, characterized in that: Also includes: The report generation module is used to automatically generate a periodic multi-disease-course health management comprehensive report based on the output of the personal digital twin module, historical first-type interaction data, and historical second-type interaction data. The report should include at least: an analysis of the changing trends of key physiological indicators, an analysis of the implementation of intervention tasks, an evolution of risk assessment across the course of the disease, and a summary of the phased management effects.

Images