An HDU patient early rehabilitation whole-process monitoring and intelligent evaluation system based on digital twinning

Abstract

The application relates to the technical field of medical assistance, and provides an HDU patient early rehabilitation whole-process monitoring and intelligent evaluation system based on digital twinning, which comprises a digital twinning model construction module, a multi-parameter monitoring module, an intelligent evaluation module, a user interaction module and a data management module. The application realizes the fundamental change of the physiological state of the HDU patient from intermittent sampling to continuous and uninterrupted panoramic perception by constructing a dynamic digital twin body that is real-time synchronized with the real patient, so that the clinical team can penetrate the appearance, intuitively understand the interaction and potential compensation mechanism inside the patient's body system, provide unprecedented depth information support for rehabilitation decision-making, and improve rehabilitation management from passive response to a new level of active perception.

Description

Technical Field

[0001] This invention relates to the field of medical assistive technology, and in particular to a digital twin-based system for monitoring and intelligently assessing the early rehabilitation process of HDU patients. Background Technology

[0002] In the traditional medical model, patients in high-dependency wards are in a critical rehabilitation window between the acute phase and the general ward. Their vital signs are relatively unstable, and rehabilitation intervention carries a high risk. Currently, monitoring of these patients mainly relies on intermittent vital sign measurements and subjective observations by medical staff. The data is fragmented and it is difficult to capture subtle dynamic changes in the condition. Rehabilitation assessments rely heavily on experience-based judgments and the phased completion of scales, which are highly subjective and lagging, and cannot achieve real-time, quantitative evaluation of functional status. This disconnect between monitoring and assessment leads to standardized but not personalized rehabilitation plans, making it difficult to accurately grasp the optimal timing and intensity of early rehabilitation. This may cause the golden window for promoting functional recovery to be missed, and it is also impossible to effectively warn of potential complications. To a certain extent, this restricts the rehabilitation efficiency and safety of HDU patients and increases the medical burden.

[0003] Therefore, in order to address the above issues, we are now developing a digital twin-based system for the whole process of early rehabilitation monitoring and intelligent assessment of HDU patients. Summary of the Invention

[0004] In order to overcome the shortcomings of existing technologies in practical use, this invention provides a digital twin-based system for monitoring and intelligently assessing the early rehabilitation process of HDU patients.

[0005] The technical solution of this invention is: a digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients, comprising: The digital twin model building module is used to create a high-fidelity digital twin model based on the individualized data of HDU patients. This model integrates the patient's electronic health records, medical imaging data, genomic information and real-time physiological parameters, and uses three-dimensional reconstruction technology and physiological simulation algorithms to construct a virtual patient entity. It can dynamically simulate the patient's cardiovascular system, respiratory system, motor function and psychological state, and predict the changing trends of key indicators during the rehabilitation process. The multi-parameter monitoring module consists of wearable sensors, bedside monitoring devices, environmental sensors, and mobile terminals. It is used to continuously collect the patient's physiological parameters, activity data, sleep quality, environmental factors, and subjective feedback throughout the entire process. Physiological parameters include heart rate, blood pressure, blood oxygen saturation, respiratory rate, and body temperature. Activity data includes steps, posture changes, and range of motion. Environmental factors include temperature, humidity, and noise level. Subjective feedback is collected through patient self-assessment scales. The monitoring data is transmitted to the central processing unit in real time via Internet of Things (IoT) technology. The intelligent assessment module employs multimodal machine learning algorithms, including deep learning models such as convolutional neural networks for feature extraction, recurrent neural networks for time series analysis, and reinforcement learning for personalized strategy optimization. This module compares and analyzes real-time monitoring data with digital twin models, calculates rehabilitation progress index, risk score, and compliance index, generates quantitative assessment reports, and provides early warnings. The user interaction module provides a graphical interface that supports access from multiple roles, including medical staff, patients, and family members. The interface displays real-time monitoring curves, assessment results, predicted trends, and rehabilitation suggestions, and integrates voice interaction and virtual reality functions to enhance the user experience. The data management module is responsible for data storage, encryption, backup, and privacy protection. It uses blockchain technology to ensure that the data is tamper-proof and complies with medical data security standards such as HIPAA.

[0006] As a preferred embodiment of the present invention, the digital twin model construction module further includes a model initialization submodule, a dynamic update submodule, and a verification submodule. The model initialization submodule, based on the patient's basic information such as age, gender, medical history, and admission diagnosis, performs organ-level three-dimensional modeling using medical imaging data such as CT and MRI, and integrates a physiological simulation engine to simulate hemodynamics to create a personalized digital twin. The dynamic update submodule uses real-time monitoring data and employs adaptive filtering algorithms such as Kalman filtering or particle filtering to adjust the digital twin parameters at a minute-level frequency to ensure that the model is synchronized with the patient's actual state and reduce prediction bias. The verification submodule evaluates the model by comparing the digital twin prediction results with actual clinical measurements using statistical methods such as root mean square error, and periodically optimizes the model accuracy. The digital twin model also includes a psychological twin submodule, which constructs a psychological state model based on the patient's behavioral data, voice analysis, and emotion questionnaires to assess psychological indicators such as anxiety and depression, achieving integrated monitoring of mind and body.

[0007] As a preferred embodiment of the present invention, the multi-parameter monitoring module includes a sensor network, a data preprocessing unit, and a real-time transmission unit. The sensor network consists of multiple types of sensors, including wearable sensors such as smart bracelets, ECG patches, and inertial measurement units, used to monitor ECG, motion, and posture. Bedside monitoring devices include multi-parameter monitors, smart mattresses, and cameras, used to collect respiratory waveforms, pressure distribution, and video data. Environmental sensors include thermometers, hygrometers, and sound sensors, used to record environmental factors. The data preprocessing unit integrates signal processing algorithms, such as wavelet transform for noise filtering, anomaly detection algorithms for identifying data anomalies, and data fusion, converting multi-source data into a standardized format. The real-time transmission unit uses 5G or Wi-Fi communication protocols to ensure low-latency data transmission and supports local preprocessing by edge computing devices to reduce cloud load. The monitoring module is also designed with a modular structure, allowing for flexible configuration of sensor types according to the patient's condition. For example, heart rate variability sensors can be prioritized for cardiovascular disease patients, while respiratory monitoring sensors can be added for patients with respiratory disorders.

[0008] As a preferred embodiment of the present invention, the intelligent assessment module includes a data fusion engine, a machine learning core, and a report generator. The data fusion engine employs multi-source information fusion technology to integrate monitoring data, digital twin output, and clinical knowledge base to generate a comprehensive feature vector. The machine learning core embeds various AI algorithms: supervised learning algorithms such as support vector machines are used for classifying the rehabilitation stage; unsupervised learning algorithms such as clustering algorithms are used for patient grouping; deep learning models such as long short-term memory networks are used to predict rehabilitation trends; and transfer learning is introduced to utilize historical data to improve assessment accuracy in small sample scenarios. The report generator automatically generates a structured report based on the assessment results, including rehabilitation scores, risk levels, recommended measures, and visualization charts. The report supports multilingual output and custom templates. The intelligent assessment module also has adaptive learning capabilities, optimizing algorithm parameters through feedback loops based on adjustments made by medical staff, and integrating expert system rules, such as clinical guideline logic, to ensure the interpretability and clinical applicability of the assessment results.

[0009] As a preferred embodiment of the present invention, the system further includes a data integration and interoperability module. This module adopts the HL7FHIR standard to achieve seamless integration with hospital information systems such as HIS and EMR, supporting real-time data exchange and interoperability. The data integration module includes an adapter layer, a conversion engine, and an API interface. The adapter layer adapts data for different device protocols, the conversion engine converts heterogeneous data into a unified format, and the API interface provides RESTful services for third-party systems to call. The interoperability module also supports real-time data stream processing, uses Apache Kafka technology to process high-concurrency data, and ensures data consistency and integrity.

[0010] As a preferred embodiment of the present invention, the user interaction module includes a multi-terminal interface, interaction logic, and a feedback mechanism. The multi-terminal interface encompasses a web terminal, a mobile app terminal, and an AR / VR terminal. The web terminal provides medical staff with a dashboard-style view, displaying a real-time monitoring panel, assessment history, and an alert list. The mobile app terminal is designed for patients and includes rehabilitation plan tracking, educational content, and instant messaging functions. The AR / VR terminal utilizes virtual reality technology to create an immersive rehabilitation training environment, such as simulating daily activities to assess functional recovery. The interaction logic adopts a user-centered design principle, integrating a voice assistant such as an NLP-based chatbot, supporting voice command queries and feedback. The feedback mechanism includes real-time alert push notifications, suggested adjustments, and satisfaction surveys. Alerts are pushed according to risk level classifications, and the interaction process is optimized through A / B testing.

[0011] As a preferred embodiment of the present invention, the system further includes a rehabilitation intervention linkage module, which is integrated with external rehabilitation equipment such as intelligent wheelchairs and robotic rehabilitation devices to achieve a closed loop of monitoring, assessment, and intervention. The rehabilitation intervention linkage module controls the rehabilitation equipment through a standard interface and automatically adjusts the equipment parameters based on the intelligent assessment results, such as adjusting the assistance level of the rehabilitation robot based on the motion assessment. The linkage module also includes a plan management submodule, which is used to generate and optimize personalized rehabilitation plans. The plans are dynamically adjusted based on the results of digital twin simulation and support manual intervention by medical staff.

[0012] As a preferred embodiment of the present invention, the system adopts a cloud-edge-device collaborative architecture. The cloud is responsible for big data storage and complex calculations, while edge nodes are deployed locally in the hospital for real-time processing, and terminal devices such as sensors perform data acquisition. The cloud-edge-device collaboration achieves resource scheduling through a distributed computing framework, ensuring low latency and high reliability. The system also supports federated learning technology, allowing hospitals to collaboratively train AI models while protecting data privacy, thereby improving the generalization ability of the evaluation models. In terms of security, the architecture includes multi-layer encryption, access control, and intrusion detection, which comply with medical network security standards.

[0013] As a preferred embodiment of the present invention, the system further includes a clinical trial verification module, which integrates a virtual trial platform and uses digital twins to simulate clinical trials, reducing the cost of actual trials; the verification module compares the system evaluation results with the gold standard clinical evaluation, such as using ROC curve analysis to assess accuracy, and generates a verification report; the module also supports long-term follow-up data collection for optimizing models and algorithms.

[0014] As a preferred embodiment of the present invention, the commercial application method of the system includes subscription services, data analysis, and customized solutions; the subscription services are charged based on the number of patients or functional modules, and provide technical support and updates; the data analysis services generate industry insight reports based on aggregated data for medical management decisions; the customized solutions are developed to meet the specific needs of hospitals; the commercial method also includes collaborating with insurance companies to optimize insurance products based on evaluation results.

[0015] By adopting the above technical solution, the present invention has the following advantages: 1. This invention achieves a fundamental shift in the perception of HDU patients' physiological state from intermittent sampling to continuous and uninterrupted panoramic perception by constructing a dynamic digital twin that is synchronized with the real patient in real time. This enables the clinical team to see beyond the surface and intuitively understand the interactions and potential compensatory mechanisms within the patient's body system, providing unprecedented in-depth information support for rehabilitation decisions and elevating rehabilitation management from a passive response to a new level of proactive perception.

[0016] 2. This invention, through its built-in multimodal artificial intelligence algorithm, can deeply mine the fused massive amounts of data, automatically identifying subtle danger signals and early deterioration trends that are difficult for humans to detect. With the help of the predictive simulation function of digital twin technology, a safety rehearsal can be conducted in a virtual space before implementing a certain rehabilitation activity, accurately assessing the stress response and tolerance limit that the patient's physiological system may produce. This shifts the intervention point forward, changing from post-event remediation to pre-event prevention, greatly reducing the medical risks in the early rehabilitation process and ensuring patient safety.

[0017] 3. This invention provides medical staff with objective decision-making basis through the generated quantitative rehabilitation index and trend report, enabling them to accurately grasp the rehabilitation pace and dynamically adjust personalized plans. The information sharing platform built by the system seamlessly connects doctors, therapists, nurses and family members, forming a multidisciplinary collaborative rehabilitation network, thereby optimizing the rehabilitation process as a whole and improving the utilization efficiency of medical resources. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0019] References to embodiments herein mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0020] A digital twin-based system for monitoring and intelligently assessing the early rehabilitation process of HDU patients, such as... Figure 1As shown, the system includes a digital twin model construction module, a multi-parameter monitoring module, an intelligent assessment module, a user interaction module, and a data management module. The digital twin model construction module creates a high-fidelity digital twin model based on individualized data from HDU patients. This model integrates the patient's electronic health records, medical imaging data, genomic information, and real-time physiological parameters, utilizing 3D reconstruction technology and physiological simulation algorithms to construct a virtual patient entity. It can dynamically simulate the patient's cardiovascular system, respiratory system, motor function, and psychological state, and predict the changing trends of key indicators during the rehabilitation process. The multi-parameter monitoring module consists of wearable sensors, bedside monitoring devices, environmental sensors, and a mobile terminal, used for continuous data collection throughout the entire process. The system collects patients' physiological parameters, activity data, sleep quality, environmental factors, and subjective feedback. Physiological parameters include heart rate, blood pressure, blood oxygen saturation, respiratory rate, and body temperature. Activity data includes steps, postural changes, and range of motion. Environmental factors include temperature, humidity, and noise levels. Subjective feedback is collected through patient self-assessment scales. Monitoring data is transmitted in real-time to the central processing unit via IoT technology. The intelligent assessment module employs multimodal machine learning algorithms, including deep learning models such as convolutional neural networks for feature extraction, recurrent neural networks for time series analysis, and reinforcement learning for personalized strategy optimization. This module compares and analyzes real-time monitoring data with a digital twin model to calculate a rehabilitation progress index and risk assessment. The system generates quantitative assessment reports and provides early warnings based on compliance indicators. The user interaction module provides a graphical interface supporting multiple roles (medical staff, patients, and family members). The interface displays real-time monitoring curves, assessment results, predicted trends, and rehabilitation suggestions, and integrates voice interaction and virtual reality functions to enhance the user experience. The data management module is responsible for data storage, encryption, backup, and privacy protection, employing blockchain technology to ensure data immutability and compliance with medical data security standards such as HIPAA. The entire system is deployed in a distributed manner through a cloud platform, supporting multi-hospital collaboration and remote monitoring, thereby achieving full-cycle rehabilitation monitoring and intelligent assessment for HDU patients from admission to discharge. The digital twin model construction module further enhances this process. The process includes a model initialization submodule, a dynamic update submodule, and a validation submodule. The model initialization submodule uses basic patient information such as age, gender, medical history, and admission diagnosis to perform organ-level 3D modeling using medical imaging data such as CT and MRI, and integrates a physiological simulation engine to simulate hemodynamics to create a personalized digital twin. The dynamic update submodule uses real-time monitoring data and adaptive filtering algorithms such as Kalman filtering or particle filtering to adjust the digital twin parameters at a frequency of minutes to ensure that the model is synchronized with the patient's actual state and reduce prediction bias. The validation submodule evaluates the model by comparing the digital twin prediction results with actual clinical measurements using statistical methods such as root mean square error, and periodically optimizes the model accuracy.The digital twin model also includes a psychological twin submodule, which constructs a psychological state model based on the patient's behavioral data, voice analysis, and emotional questionnaires. This model is used to assess psychological indicators such as anxiety and depression, achieving integrated mind-body monitoring. The multi-parameter monitoring module includes a sensor network, a data preprocessing unit, and a real-time transmission unit. The sensor network consists of various types of sensors, including wearable sensors such as smart bracelets, ECG patches, and inertial measurement units, used to monitor ECG, movement, and posture. Bedside monitoring devices include multi-parameter monitors, smart mattresses, and cameras, used to collect respiratory waveforms, pressure distribution, and video data. Environmental sensors include thermometers, hygrometers, and sound sensors, used to record environmental factors. The data preprocessing unit integrates signal processing algorithms, such as wavelet transform. The noise filtering and anomaly detection algorithms are used to identify data anomalies and achieve data fusion, converting multi-source data into a standardized format. The real-time transmission unit uses 5G or Wi-Fi communication protocols to ensure low-latency data transmission and supports local preprocessing on edge computing devices to reduce cloud load. The monitoring module is also designed with a modular structure, allowing for flexible configuration of sensor types according to the patient's condition. For example, heart rate variability sensors can be prioritized for cardiovascular disease patients, and respiratory monitoring sensors can be added for patients with respiratory disorders. The intelligent assessment module includes a data fusion engine, a machine learning core, and a report generator. The data fusion engine uses multi-source information fusion technology to integrate monitoring data, digital twin output, and clinical knowledge base to generate a comprehensive feature vector. The core of the system incorporates multiple AI algorithms: supervised learning algorithms such as support vector machines for classifying the rehabilitation phase, unsupervised learning algorithms such as clustering for patient grouping, and deep learning models such as long short-term memory networks for predicting rehabilitation trends. Transfer learning is also introduced to leverage historical data to improve assessment accuracy in small-sample scenarios. The report generator automatically generates structured reports based on assessment results, including rehabilitation scores, risk levels, recommended measures, and visualization charts. The reports support multilingual output and custom templates. The intelligent assessment module also features adaptive learning capabilities, optimizing algorithm parameters through feedback loops based on adjustments made by medical staff, and integrating expert system rules, such as clinical guideline logic, to ensure the interpretability and clinical applicability of assessment results. The user interaction module includes a multi-terminal interface. The system encompasses a multi-platform interface including web, mobile app, and AR / VR. The web interface provides healthcare professionals with a dashboard view displaying real-time monitoring panels, assessment history, and alert lists. The mobile app is designed for patients, including rehabilitation plan tracking, educational content, and instant messaging functionality. The AR / VR platform utilizes virtual reality technology to create immersive rehabilitation training environments, such as simulating daily activities to assess functional recovery. The interaction logic adopts a user-centered design principle, integrating voice assistants such as NLP-based chatbots to support voice command queries and feedback. The feedback mechanism includes real-time alert push notifications, suggested adjustments, and satisfaction surveys. Alerts are pushed according to risk level grading, and the interaction process is optimized through A / B testing.

[0021] It also includes a data integration and interoperability module, which uses the HL7 FHIR standard to achieve seamless integration with hospital information systems such as HIS and EMR, supporting real-time data exchange and interoperability. The data integration module includes an adapter layer, a transformation engine, and an API interface. The adapter layer adapts data for different device protocols, the transformation engine converts heterogeneous data into a unified format, and the API interface provides RESTful services for third-party systems to call. The interoperability module also supports real-time data stream processing, using Apache Kafka technology to handle high-concurrency data and ensure data consistency and integrity.

[0022] It also includes a rehabilitation intervention linkage module, which integrates with external rehabilitation equipment such as smart wheelchairs and robotic rehabilitation devices to achieve a closed loop of monitoring, assessment, and intervention. The rehabilitation intervention linkage module controls the rehabilitation equipment through a standard interface and automatically adjusts the equipment parameters based on the intelligent assessment results, such as adjusting the level of assistance of the rehabilitation robot based on the motion assessment. The linkage module also includes a plan management sub-module, which is used to generate and optimize personalized rehabilitation plans. The plans are dynamically adjusted based on the results of digital twin simulation and support manual intervention by medical staff.

[0023] It also includes a clinical trial validation module, which integrates a virtual trial platform to simulate clinical trials using digital twins, reducing the cost of actual trials; the validation module compares the results of system evaluation with the gold standard clinical evaluation, such as using ROC curve analysis to assess accuracy, and generates validation reports; the module also supports long-term follow-up data collection for optimizing models and algorithms.

[0024] The system adopts a cloud-edge-device collaborative architecture. The cloud is responsible for big data storage and complex calculations, while edge nodes are deployed locally in the hospital for real-time processing. Terminal devices, such as sensors, perform data acquisition. The cloud-edge-device collaboration achieves resource scheduling through a distributed computing framework, ensuring low latency and high reliability. The system also supports federated learning technology, allowing hospitals to collaboratively train AI models while protecting data privacy, thereby improving the generalization ability of the evaluation models. In terms of security, the architecture includes multi-layered encryption, access control, and intrusion detection, complying with medical network security standards. The commercial application methods of the system include subscription services, data analysis, and customized solutions. Subscription services are charged based on the number of patients or functional modules, providing technical support and updates. Data analysis services generate industry insight reports based on aggregated data for medical management decisions. Customized solutions are developed to meet the specific needs of hospitals. The commercial approach also includes cooperation with insurance companies to optimize insurance products based on evaluation results.

[0025] It should be noted that the system is activated immediately upon the admission of a patient who has just passed the acute critical period and been transferred from the ICU to the HDU. Specifically, once the patient is admitted to the HDU bed, an integrated multi-parameter monitoring network automatically starts. This network includes a multi-lead smart patch attached to the patient's chest to continuously monitor ECG, respiratory waveforms, and body temperature; a medical-grade smart bracelet worn on the wrist to monitor blood oxygen saturation and activity level; and a smart bed with built-in distributed pressure sensors that can subtly detect changes in the patient's position, the number of times they turn over, and even muscle tension. At the same time, a high-definition camera next to the bed (with privacy protection measures) uses computer vision algorithms to non-contactly monitor respiratory rate and limb range of motion, as well as temperature, humidity, light, and noise levels in the ward environment. Sensors record external factors that may affect patient recovery. All this monitoring data, including high-frequency physiological waveforms and low-frequency environmental data, undergoes preliminary preprocessing via edge computing gateways deployed near the bedside. This preprocessing includes using wavelet transform algorithms to filter motion artifacts and power frequency interference, ensuring data quality. The data is then transmitted in real-time to the central data processing platform via the hospital's 5G or Wi-Fi network in a low-latency, highly reliable manner. In the platform's data management module, the massive influx of multi-source heterogeneous data is first standardized, timestamped, and tagged with patient IDs. It is then stored in a time-series database and a non-relational database, and hash verification is performed using blockchain technology to ensure the integrity and immutability of the data during transmission and storage. Adhering to the highest standards of medical data security, the digital twin model building module is activated. It first retrieves the patient's baseline data from the Hospital Information System (HIS), Electronic Medical Record (EMR), and Picture Archiving System (PACS), including age, gender, admission diagnosis (e.g., post-acute myocardial infarction surgery), past medical history, laboratory test results, and cardiac CT images. Based on this static data, the model initialization submodule uses 3D reconstruction technology to generate precise geometric models of the patient's heart, lungs, and other key organs based on the CT data. This is then coupled with a physiological simulation engine, such as using computational fluid dynamics to simulate the patient's coronary artery blood flow or using a physics-based model to simulate respiratory mechanics, thereby constructing a digital twin model that includes… A personalized digital twin based on anatomical structure and physiological function is created. Then, a dynamic update submodule begins operating. It compares the continuously incoming real-time monitoring data (such as heart rate, blood oxygen, and respiratory curves per second) with the predicted output of the digital twin model. Adaptive algorithms, such as Kalman filtering, are used for data assimilation. This dynamically corrects and updates thousands of parameters within the digital twin model (such as vascular resistance, lung compliance, and myocardial contractility) at a frequency of minutes or even seconds. This ensures that the virtual model remains highly synchronized with the physiological state of the real patient. When healthcare professionals plan to perform an early bedside sitting-up exercise for the patient, they can input the rehabilitation command "Plan 30 minutes of 60-degree sitting training" in the system's user interface.The intelligent assessment module immediately drives the digital twin model to perform a pre-training exercise. Based on the patient's current physiological state (such as a slightly faster heart rate and higher blood pressure), and inputting the stress conditions for sit-up training, the module rapidly simulates potential changes in the patient's cardiovascular system over the next 30 minutes using built-in physiological equations. It predicts trends in cardiac output, fluctuations in blood pressure, and potential risks to myocardial oxygen consumption, providing a quantified risk score. If the prediction results indicate that a certain indicator (such as a predicted decrease in systolic blood pressure exceeding 30 mmHg) may exceed a safe threshold, the system immediately pushes a yellow or red alert to the nurses' station's smart screen and the medical staff's mobile PDAs via the user interaction module, suggesting adjusting the training angle to 45 degrees or increasing the blood pressure during training. Oxygen monitoring allows for safety assessment and optimization in a virtual space before implementing rehabilitation measures in the physical world, enabling proactive intervention and significantly reducing rehabilitation risks. Beyond predictive simulation, the core function of the intelligent assessment module is continuous and objective evaluation based on the fusion of real and digital twin data. This module incorporates multimodal machine learning algorithms, such as using convolutional neural networks to automatically extract subtle ST segment changes from ECG signals and recurrent neural networks to analyze time-series patterns of heart rate variability to assess the recovery of autonomic nervous function. Simultaneously, it continuously compares real-time patient activity data (such as total daily activity and time out of bed calculated from the fusion of smart bracelet and mattress pressure sensors) with the digital twin based on the ideal rehabilitation path. The module calculates a daily rehabilitation progress index by comparing the predicted activity levels with the actual activity levels. It also integrates data from short psychological scales (such as the Hospital Anxiety and Depression Scale) completed daily by patients via tablets, and even analyzes acoustic features from their voice recordings. A sub-model of the mental state is constructed in the digital twin to assess the risk of anxiety and depression. All these physiological, functional, and psychological indicators are integrated by a data fusion engine based on DS evidence theory or Bayesian networks, ultimately generating a comprehensive, quantitative rehabilitation assessment report. This report not only includes a current status score but also predicts the rehabilitation trend for the next three days using long short-term memory networks, clearly identifying bottlenecks in the rehabilitation process (such as delayed muscle strength recovery) or potential complications (such as the risk of deep vein thrombosis). The system adds personalized rehabilitation plan adjustments, such as suggesting increasing lower limb active joint range of motion training to three times daily. Finally, the system's user interaction module presents all this complex information in an intuitive way to users in different roles. For doctors and rehabilitation therapists, a 3D visualization model of the patient's digital twin is displayed on a smart screen in the ward or on their computer terminal, with key physiological parameters overlaid with dynamic curves, alongside real-time assessment scores and warning information. They can click on the model to rotate and zoom, view the simulated state of specific organs, and drag and drop different rehabilitation programs on the twin model to assist clinical decision-making. For the patient, a simple rehabilitation game interface is presented on a bedside tablet.The virtual avatar's movements are directly driven by the patient's real activities, and completing training tasks earns rewards, enhancing the fun and adherence to rehabilitation. For family members, an authorized mobile app allows them to securely view the patient's daily rehabilitation summaries and trend charts, alleviating their anxiety. The entire system utilizes a cloud-edge-device collaborative architecture to ensure efficient and secure data processing. All assessment data, early warning records, and medical feedback form a closed loop, continuously optimizing the accuracy of the machine learning model and digital twin. Thus, in the critical HDU rehabilitation stage, it provides unprecedented technological support for achieving safe, efficient, and personalized early rehabilitation for patients.

[0026] The above embodiments are provided for those skilled in the art to implement or use the present invention. Those skilled in the art can make various modifications or changes to the above embodiments without departing from the spirit of the present invention. Therefore, the scope of protection of the present invention is not limited to the above embodiments, but should be the maximum scope that conforms to the innovative features mentioned in the claims.

Claims

1. A digital twin-based system for monitoring and intelligently assessing the early rehabilitation process of HDU patients, characterized in that, Including: The digital twin model building module is used to create a high-fidelity digital twin model based on the individualized data of HDU patients. This model integrates the patient's electronic health records, medical imaging data, genomic information and real-time physiological parameters, and uses three-dimensional reconstruction technology and physiological simulation algorithms to construct a virtual patient entity. It can dynamically simulate the patient's cardiovascular system, respiratory system, motor function and psychological state, and predict the changing trends of key indicators during the rehabilitation process. The multi-parameter monitoring module consists of wearable sensors, bedside monitoring devices, environmental sensors, and mobile terminals. It is used to continuously collect the patient's physiological parameters, activity data, sleep quality, environmental factors, and subjective feedback throughout the entire process. Physiological parameters include heart rate, blood pressure, blood oxygen saturation, respiratory rate, and body temperature. Activity data includes steps, posture changes, and range of motion. Environmental factors include temperature, humidity, and noise level. Subjective feedback is collected through patient self-assessment scales. The monitoring data is transmitted to the central processing unit in real time via Internet of Things (IoT) technology. The intelligent assessment module employs multimodal machine learning algorithms, including deep learning models such as convolutional neural networks for feature extraction, recurrent neural networks for time series analysis, and reinforcement learning for personalized strategy optimization. This module compares and analyzes real-time monitoring data with digital twin models, calculates rehabilitation progress index, risk score, and compliance index, generates quantitative assessment reports, and provides early warnings. The user interaction module provides a graphical interface that supports access from multiple roles, including medical staff, patients, and family members. The interface displays real-time monitoring curves, assessment results, predicted trends, and rehabilitation suggestions, and integrates voice interaction and virtual reality functions to enhance the user experience. The data management module is responsible for data storage, encryption, backup, and privacy protection. It uses blockchain technology to ensure that the data is tamper-proof and complies with medical data security standards such as HIPAA.

2. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The digital twin model construction module further includes a model initialization submodule, a dynamic update submodule, and a validation submodule. The model initialization submodule, based on the patient's basic information such as age, gender, medical history, and admission diagnosis, performs organ-level 3D modeling using medical imaging data such as CT and MRI, and integrates a physiological simulation engine to simulate hemodynamics to create a personalized digital twin. The dynamic update submodule uses real-time monitoring data and employs adaptive filtering algorithms such as Kalman filtering or particle filtering to adjust the digital twin parameters at a minute-level frequency, ensuring the model is synchronized with the patient's actual state and reducing prediction bias. The validation submodule evaluates the model by comparing the digital twin prediction results with actual clinical measurements using statistical methods such as root mean square error, and periodically optimizes the model's accuracy. The digital twin model also includes a psychological twin submodule, which constructs a psychological state model based on the patient's behavioral data, voice analysis, and emotional questionnaires to assess psychological indicators such as anxiety and depression, achieving integrated mind-body monitoring.

3. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The multi-parameter monitoring module includes a sensor network, a data preprocessing unit, and a real-time transmission unit. The sensor network consists of various types of sensors, including wearable sensors such as smart bracelets, ECG patches, and inertial measurement units, used to monitor ECG, motion, and posture. Bedside monitoring devices include multi-parameter monitors, smart mattresses, and cameras, used to collect respiratory waveforms, pressure distribution, and video data. Environmental sensors include thermometers, hygrometers, and sound sensors, used to record environmental factors. The data preprocessing unit integrates signal processing algorithms, such as wavelet transform for noise filtering, anomaly detection algorithms for identifying data anomalies, and achieves data fusion, converting multi-source data into a standardized format. The real-time transmission unit uses 5G or Wi-Fi communication protocols to ensure low-latency data transmission and supports local preprocessing by edge computing devices to reduce cloud load. The monitoring module is also designed with a modular structure, which can flexibly configure the sensor type according to the patient's condition. For example, heart rate variability sensors can be prioritized for cardiovascular disease patients, and respiratory monitoring sensors can be added for patients with respiratory disorders.

4. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The intelligent assessment module includes a data fusion engine, a machine learning core, and a report generator; The data fusion engine uses multi-source information fusion technology to integrate monitoring data, digital twin output, and clinical knowledge base to generate a comprehensive feature vector; The core of machine learning incorporates a variety of AI algorithms, including supervised learning algorithms such as support vector machines for classifying the rehabilitation stage, unsupervised learning algorithms such as clustering algorithms for patient grouping, and deep learning models such as long short-term memory networks for predicting rehabilitation trends. Transfer learning is also introduced to improve the assessment accuracy in small sample scenarios by utilizing historical data. The report generator automatically generates structured reports based on assessment results, including rehabilitation scores, risk levels, recommended measures, and visualization charts. The reports support multilingual output and custom templates. The intelligent assessment module also has adaptive learning capabilities, optimizing algorithm parameters through feedback loops based on adjustments made by healthcare professionals, and integrating expert system rules, such as clinical guideline logic, to ensure the interpretability and clinical applicability of assessment results.

5. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The system also includes a data integration and interoperability module, which uses the HL7 FHIR standard to achieve seamless integration with hospital information systems such as HIS and EMR, and supports real-time data exchange and interoperability. The data integration module includes an adapter layer, a conversion engine, and an API interface. The adapter layer adapts data for different device protocols, the conversion engine converts heterogeneous data into a unified format, and the API interface provides RESTful services for third-party systems to call. The interoperability module also supports real-time data stream processing, using Apache Kafka technology to handle high-concurrency data and ensure data consistency and integrity.

6. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The user interaction module includes a multi-terminal interface, interaction logic, and feedback mechanism. The multi-terminal interface covers a web terminal, a mobile app terminal, and an AR / VR terminal. The web terminal provides medical staff with a dashboard-style view, displaying a real-time monitoring panel, assessment history, and an early warning list. The mobile app is designed for patients and includes rehabilitation plan tracking, educational content, and instant messaging functions; the AR / VR app uses virtual reality technology to create an immersive rehabilitation training environment, such as simulating daily activities to assess functional recovery. The interaction logic adopts a user-centered design principle, integrating voice assistants such as NLP-based chatbots, and supports voice command queries and feedback; The feedback mechanism includes real-time alert push notifications, suggested adjustments, and satisfaction surveys. Alerts are pushed out according to risk level classifications, and the interaction process is optimized through A / B testing.

7. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The system also includes a rehabilitation intervention linkage module, which integrates with external rehabilitation equipment such as smart wheelchairs and robotic rehabilitation devices to achieve a closed loop of monitoring, assessment, and intervention. The rehabilitation intervention linkage module controls the rehabilitation equipment through a standard interface and automatically adjusts the equipment parameters based on intelligent assessment results, such as adjusting the level of assistance from the rehabilitation robot based on motion assessment. The linkage module also includes a plan management submodule, which is used to generate and optimize personalized rehabilitation plans. The plans are dynamically adjusted based on digital twin simulation results and support manual intervention by medical staff.

8. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The system adopts a cloud-edge-device collaborative architecture. The cloud is responsible for big data storage and complex calculations, while edge nodes are deployed locally in the hospital for real-time processing. Terminal devices, such as sensors, perform data acquisition. The cloud-edge-device collaboration achieves resource scheduling through a distributed computing framework, ensuring low latency and high reliability. The system also supports federated learning technology, allowing hospitals to collaboratively train AI models while protecting data privacy, thereby improving the generalization ability of the evaluation models. In terms of security, the architecture includes multi-layer encryption, access control, and intrusion detection, complying with medical network security standards.

9. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The system also includes a clinical trial validation module, which integrates a virtual trial platform and uses digital twins to simulate clinical trials, reducing the cost of actual trials. The validation module compares the system evaluation results with the gold standard clinical evaluation, such as using ROC curve analysis to assess accuracy, and generates a validation report. The module also supports long-term follow-up data collection for optimizing models and algorithms.

10. A digital twin-based early rehabilitation monitoring and intelligent assessment system for HDU patients according to claim 1, characterized in that, The commercial application methods of the system include subscription services, data analysis, and customized solutions; subscription services are charged based on the number of patients or functional modules, and provide technical support and updates; Data analytics services generate industry insight reports based on aggregated data for healthcare management decision-making; customized solutions are developed to meet the specific needs of hospitals. The business approach also includes collaborating with insurance companies to optimize insurance products based on assessment results.

Images