A hemodialysis patient rehabilitation exercise auxiliary decision system and method
By constructing a rehabilitation exercise auxiliary decision-making system that integrates multi-source data acquisition, intelligent analysis, and decision-making, the system solves the problem of personalized decision-making for hemodialysis patients during exercise, achieving precise rehabilitation and improved safety, and empowering nursing staff to manage more patients.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 厦门市第五医院(厦门市同民医院)
- Filing Date
- 2026-03-09
- Publication Date
- 2026-06-05
AI Technical Summary
Hemodialysis patients face challenges such as insufficient understanding, lack of motivation, data silos, and high risks during rehabilitation exercises. Existing technologies struggle to provide personalized, safe, and dynamic exercise support decisions.
By employing a multi-source data acquisition layer, an intelligent analysis and decision-making layer, and an interactive execution layer, combined with a feedback optimization closed loop, a personalized rehabilitation exercise auxiliary decision-making system is constructed. AI models are used to generate precise exercise prescriptions, and prescription strategies are optimized through real-time monitoring and feedback.
It enables personalized decision-making, improves rehabilitation outcomes and efficiency, ensures safety, reduces the risk of adverse events, enhances scientific rigor and patient compliance, and empowers clinical nursing.
Smart Images
Figure CN122157961A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical monitoring technology, and in particular to a decision-making system and method for rehabilitation exercise of hemodialysis patients. Background Technology
[0002] Hemodialysis is the primary replacement therapy for patients with end-stage renal disease. However, patients commonly experience muscle atrophy, decreased physical strength, fatigue, impaired cardiovascular function, depression, and a low quality of life, a condition known as "dialysis frailty syndrome." Regular and scientific rehabilitation exercises have been proven by numerous studies to significantly improve the physiological function, cardiopulmonary endurance, nutritional status, psychological state, and overall quality of life of dialysis patients. However, the promotion of rehabilitation exercises for hemodialysis patients faces significant challenges in clinical practice: From the patient's perspective: insufficient understanding, strong fatigue, lack of motivation and continuous guidance; From the medical perspective: patients often have multiple complications such as hypertension, heart failure, anemia, and bone mineral metabolism disorders, exercise prescriptions need to be highly individualized, which is extremely risky and data silos exist. Patients' vital signs, dialysis parameters, laboratory indicators, subjective feelings and other information are scattered and cannot be effectively integrated for dynamic assessment of exercise safety and effectiveness.
[0003] Therefore, there is an urgent need for a motion-assisted decision-making tool that can integrate multi-source data, conduct personalized risk assessments, and generate accurate, safe, and dynamically adjustable data. Summary of the Invention
[0004] The purpose of this invention is to provide a decision-making system and method for rehabilitation exercise of hemodialysis patients.
[0005] To achieve the objectives of this invention, the present invention is implemented through the following technical solution: a rehabilitation exercise auxiliary decision-making system for hemodialysis patients, comprising: The data acquisition layer is used to acquire and fuse patient data from multiple sources; The intelligent analysis and decision-making layer is used to receive patient data and build a multimodal risk assessment AI model for patients to generate personalized exercise prescriptions; The interactive execution layer is used to display the patient's exercise risk, provide exercise guidance, real-time monitoring, and feedback to the patient; The feedback optimization closed loop is used to continuously collect exercise execution data, physiological responses and periodic assessment results, and feed them back to the AI model to continuously iterate and optimize the prescription generation strategy for the patient. The data acquisition layer, intelligent analysis and decision-making layer, interactive execution layer, and feedback optimization closed loop are interconnected.
[0006] Preferably, the data acquisition layer includes a hemodialysis machine interface, a wearable device, a biochemical and medical record interface, and an electronic scale.
[0007] Preferably, the hemodialysis machine interface is connected to the hemodialysis machine to acquire real-time time-series data such as the patient's heart rate, blood pressure, blood oxygen saturation, and ultrafiltration volume / rate. The wearable device includes one of a smart bracelet and a chest patch. The biochemical and medical record interface is connected to the hospital information system and the laboratory information system to acquire key laboratory indicators such as dry weight, hemoglobin, serum potassium, and troponin, as well as diagnostic and medication history. The electronic scale guides patients to regularly fill out fatigue scales, depression and anxiety scales, quality of life scales, and subjective motor experience scales via a tablet computer or mobile APP.
[0008] Preferably, the intelligent analysis and decision-making layer includes a patient digital profiling module, a hybrid risk assessment model, and a prescription generation model.
[0009] Preferably, the patient profiling module is used to integrate all static and dynamic data to construct a dynamic digital profile for each patient, including underlying diseases, physical fitness level, risk labels, and exercise history. The hybrid risk assessment model adopts a hybrid model that combines rules and data-driven approaches. The rule base of the hybrid risk assessment model includes exercise contraindication rules based at least on real-time blood pressure changes, blood potassium concentration, and ultrafiltration rate. The prescription generation model adopts a reinforcement learning algorithm, using periodic patient functional assessment results as reward signals to dynamically adjust exercise prescription parameters.
[0010] Preferably, the interaction and execution layer includes a nurse console and a patient app. The nurse console is used to display an overview of the movement risks of all patients in the ward, individualized exercise prescription suggestions, and real-time alarm information. The patient app is used for patient exercise guidance, real-time monitoring and feedback, motivation and education.
[0011] This invention also provides a method for generating personalized exercise prescriptions for hemodialysis patients, comprising the following steps: S1: acquiring and fusing multi-source patient data; S2: running a hybrid risk assessment model; S3: if the risk is acceptable, calling an AI prescription model to generate a recommended plan; S4: outputting and executing the plan through the human-computer interface in the human-computer interaction module.
[0012] Preferably, the human-computer interaction module enables real-time two-way communication of physiological data and automatic intervention logic during the patient's movement.
[0013] The present invention has the following advantages: (1) Achieve personalized decision-making: By using artificial intelligence technology, the patient's multi-dimensional static and dynamic data are integrated to generate a precise and fully personalized rehabilitation exercise prescription, thereby improving rehabilitation effect and efficiency and achieving precise rehabilitation; (2) Achieve dynamic safety early warning: Establish a real-time risk monitoring and early warning model, and adjust or terminate exercise recommendations in real time based on the patient's physiological feedback before and during exercise to ensure patient safety, greatly improve the safety and effectiveness of exercise, and significantly reduce the risk of exercise-related adverse events; (3) Achieve closed-loop management: Construct a digital closed loop of "assessment-decision-execution-feedback-re-optimization" to improve the scientific nature, safety and patient compliance of sports rehabilitation; (4) Empowering clinical nursing: Providing hemodialysis nurses with an efficient and precise intelligent auxiliary tool, freeing them from heavy experience-based judgment, allowing them to focus on humanistic care and key interventions, and enabling them to manage more patients. Attached Figure Description
[0014] Figure 1 This is a schematic diagram of the overall system framework of the present invention; Figure 2 This is a flowchart illustrating the hybrid risk assessment and decision generation process of this invention. Figure 3 This is a schematic diagram of the nurse control console interface of the present invention; Figure 4 This is a flowchart illustrating the steps of the method for assisting decision-making in rehabilitation exercise for hemodialysis patients according to the present invention. Detailed Implementation
[0015] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments and accompanying drawings.
[0016] In the description of this invention, the terms "upper," "lower," "left," "right," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.
[0017] In the description of this invention, it should be understood that the terms “comprising” and “having” as used herein, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.
[0018] like Figure 1-3The system shown is a rehabilitation exercise auxiliary decision-making system for hemodialysis patients, comprising: a data acquisition layer, an intelligent analysis and decision-making layer, an interactive execution layer, and a feedback optimization closed loop, wherein the data acquisition layer, the intelligent analysis and decision-making layer, the interactive execution layer, and the feedback optimization closed loop are interconnected.
[0019] The data acquisition layer is used to acquire and integrate multi-source patient data. The data acquisition layer includes a hemodialysis machine interface, wearable devices, a biochemical and medical record interface, and an electronic scale. The hemodialysis machine interface is connected to the hemodialysis machine and is used to acquire real-time time-series data such as the patient's heart rate, blood pressure, blood oxygen saturation, and ultrafiltration volume / rate. The wearable device includes either a smart bracelet or a chest patch. The biochemical and medical record interface is connected to the hospital information system and the laboratory information system and is used to acquire key laboratory indicators such as dry weight, hemoglobin, serum potassium, and troponin, as well as diagnosis and medication history. The electronic scale guides patients to regularly fill out fatigue scales, depression and anxiety scales, quality of life scales, and subjective motor experience scales via a tablet computer or mobile APP. The intelligent analysis and decision-making layer includes a patient digital profiling module, a hybrid risk assessment model, and a prescription generation model. The patient profiling module integrates all static and dynamic data to construct a dynamic digital profile for each patient, including underlying diseases, fitness level, risk tags, and exercise history. The hybrid risk assessment model adopts a hybrid model that combines rules and data-driven approaches. The rule base of the hybrid risk assessment model includes exercise contraindication rules based at least on real-time blood pressure changes, blood potassium concentration, and ultrafiltration rate. The prescription generation model uses a reinforcement learning algorithm, with periodic patient functional assessment results as reward signals, to dynamically adjust exercise prescription parameters. The interaction and execution layer includes a nurse console and a patient app. The nurse console is used to display an overview of the movement risks of all patients in the ward, individualized exercise prescription suggestions, and real-time alarm information. The patient app is used for patient exercise guidance, real-time monitoring and feedback, motivation and education.
[0020] It should be noted that the data acquisition layer includes a hemodialysis machine interface, wearable devices, biochemical and medical record interfaces, and electronic scales. It is used to acquire all static and dynamic data of the patient in real time. By collecting multi-source data and utilizing multi-source heterogeneous data fusion methods in the dialysis scenario, it performs time alignment, normalization, and feature engineering on time-series dialysis treatment data, intermittent laboratory data, continuous wearable device data, and discrete patient subjective reports to construct a unified patient state vector. This allows for comprehensive monitoring of the patient's vital signs and other data. Subsequently, the data acquisition layer transmits the constructed unified patient state vector to the intelligent analysis and decision-making layer. The intelligent analysis and decision-making layer receives patient data and establishes a multimodal risk assessment AI model for the patient, generating personalized exercise prescriptions. The intelligent analysis and decision-making layer includes... The system includes a patient digital profiling module, a hybrid risk assessment model, and a prescription generation model. The patient profiling module integrates all static and dynamic data collected from the data acquisition layer and constructs a dynamic digital profile for each patient based on the integrated data. This profile includes the patient's underlying diseases, fitness level, risk tags, and exercise history. The hybrid risk assessment model is built based on input real-time vital signs (such as HR, BP, and SpO2 trends), laboratory alarm values (such as hyperkalemia), dialysis parameters (such as excessive ultrafiltration), and subjective reports (such as severe fatigue). This hybrid risk assessment model employs a combination of rule-based and data-driven approaches. In this embodiment, the risk assessment model is not simply a data-driven model, but rather uses a "rule engine (hard safety rules) +..." A hybrid risk assessment model is generated using a combination of AI models (flexible optimization suggestions) and other technologies. This complementary approach balances system safety, intelligence, and maintainability. The hybrid risk assessment model's rule base includes exercise contraindications based on real-time blood pressure changes, serum potassium levels, and ultrafiltration rate. Furthermore, the intelligent analysis and decision-making layer can generate a model with real-time risk levels and key risk factor indicators based on these input features and output it to the interactive execution layer. For example, green indicates "safe," yellow indicates "cautious," and red indicates "prohibited," thus providing risk warnings to patients. Subsequently, the intelligent decision analysis layer... The constructed hybrid risk assessment model generates corresponding personalized prescriptions for patients, enabling personalized decision-making. By generating personalized prescriptions for patients, it ensures that prescriptions are generated according to the different conditions of each patient. By utilizing artificial intelligence technology, multi-dimensional static and dynamic data of patients are integrated to generate accurate and fully personalized rehabilitation exercise prescriptions, which is conducive to improving rehabilitation effects and efficiency and achieving precise rehabilitation. In other embodiments, a Bayesian network can also be used to replace the AI model part in the hybrid model. Bayesian networks can handle uncertain reasoning well and visualize the probabilistic relationship between symptoms, signs, and risks, which is very suitable for medical decision-making.The prescription generation model is based on reinforcement learning or decision tree models, aiming to improve patient functional indicators. Combining current digital profiles and risk levels, it recommends optimal exercise programs from an exercise library, accurately recommending exercise type, intensity, duration, frequency, and weekly progression rules. The exercise library includes various types of bedside exercises during dialysis, such as resistance band exercises, cycling, and interdialysis aerobic exercises like walking. An interactive execution layer displays patient exercise risks, provides exercise guidance, real-time monitoring, and feedback. This layer includes a nurse console and a patient app. The nurse console displays an overview of patient exercise risks across the entire ward, individualized exercise prescription suggestions, and real-time alarm information. Nurses can review, fine-tune, and issue prescriptions to patients with a single click. The patient app provides text, images, and videos. The system guides patients through their current exercise prescription, while a patient app provides real-time monitoring and feedback. During exercise, the app receives data from wearable devices; if the heart rate exceeds a safe range or the patient experiences discomfort, it issues an alert to immediately stop exercise and notifies the nursing station. Furthermore, the app provides motivation and education, recording exercise achievements, offering points rewards, and pushing health knowledge. It treats restricted exercise during dialysis treatment and free exercise at home as a whole for prescription design and effect tracking. By establishing a real-time risk monitoring and early warning model, it adjusts or terminates exercise recommendations based on the patient's physiological feedback before and during exercise, ensuring patient safety and greatly improving the safety and effectiveness of exercise, significantly reducing the risk of exercise-related adverse events. In addition, by setting up a feedback optimization closed loop, the system continuously collects exercise execution data, physiological responses, and periodic evaluation results, feeding them back to the AI model to iteratively optimize the prescription generation strategy for the patient, achieving increasing accuracy with use. By constructing a digital closed loop of "assessment-decision-execution-feedback-re-optimization," the system improves the scientific nature, safety, and patient compliance of exercise rehabilitation.
[0021] In summary, this invention utilizes artificial intelligence technology to integrate multi-dimensional static and dynamic data from patients, generating precise and fully personalized rehabilitation exercise prescriptions to improve rehabilitation effectiveness and efficiency, achieving precise rehabilitation. It establishes a real-time risk monitoring and early warning model, adjusting or terminating exercise recommendations based on patient physiological feedback before and during exercise, ensuring patient safety, greatly improving the safety and effectiveness of exercise, and significantly reducing the risk of exercise-related adverse events. It constructs a digital closed loop of "assessment-decision-execution-feedback-re-optimization," enhancing the scientific rigor, safety, and patient compliance of exercise rehabilitation. Furthermore, it provides hemodialysis nurses with an efficient and precise intelligent auxiliary tool, freeing them from heavy experiential judgment and allowing them to focus on humanistic care and key interventions, enabling them to manage more patients.
[0022] like Figure 4As shown, the present invention also provides a method for assisting decision-making in rehabilitation exercise for hemodialysis patients, including the following steps: S1: acquiring and fusing multi-source patient data; S2: running a hybrid risk assessment model; S3: if the risk is acceptable, calling an AI prescription model to generate a recommended plan; S4: outputting and executing the plan through the human-computer interaction module's human-computer interface, wherein the human-computer interaction module realizes real-time two-way communication of physiological data and automatic intervention logic during the patient's exercise.
[0023] The specific implementation method for assisting decision-making in rehabilitation exercises for hemodialysis patients is as follows: Example: Patient Chen, male, 58 years old, has been on maintenance hemodialysis for 3 years and has hypertension. Today, before dialysis, his serum potassium was 5.6 mol / L and his hemoglobin was 105 g / L.
[0024] Dialysis begins: The system retrieves the latest test data from HIS / LIS and marks "high blood potassium" as a warning.
[0025] Pre-exercise assessment (1 hour after dialysis begins, vital signs are stable): The nurse selects Mr. Chen on the control panel, and the system automatically runs the assessment.
[0026] Data fusion: Current BP 145 / 88 mmHg, HR 78 bpm, with "high potassium" label overlaid.
[0027] Risk assessment: The rule engine determined that blood pressure and heart rate were within safe ranges, but due to potassium levels >5.5, the risk level was set to "Yellow - Caution". The AI model analysis showed that the individual had a good history of exercise tolerance.
[0028] Prescription generation: Under the premise of "caution", the AI model recommends a set of low-to-moderate intensity resistance exercises for dialysis, focusing on the lower limbs (to avoid the risk of strenuous upper limb exercise due to high potassium), and reduces the intensity of the original plan by 10%.
[0029] Exercise execution and monitoring: The prescription was sent to a tablet next to Chen's bed, which played video instructions such as ankle pumps with elastic bands and straight leg raises.
[0030] During the execution, Chen's heart rate data was transmitted in real time. The system monitored his heart rate as it steadily rose to the lower limit of the target heart rate range.
[0031] Emergency Simulation: If the system suddenly receives an alarm from the hemodialysis machine indicating "abnormal fluctuation in arterial pressure" or if Mr. Chen clicks "feels dizzy", the real-time monitoring module will immediately trigger a "pause exercise" command, the tablet will display a stop, and an alarm will be simultaneously sent to the nurse station, where a nurse will immediately arrive to handle the situation.
[0032] Feedback and Recording: The exercise was successfully completed, and Mr. Chen rated his RPE as "a little tired." The system recorded 100% exercise adherence and average heart rate response. This data will be incorporated into his personal file for future prescription optimization.
[0033] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.
Claims
1. A rehabilitation exercise auxiliary decision-making system for hemodialysis patients, characterized in that, include: The data acquisition layer is used to acquire and fuse patient data from multiple sources; The intelligent analysis and decision-making layer is used to receive patient data and build a multimodal risk assessment AI model for patients to generate personalized exercise prescriptions; The interactive execution layer is used to display the patient's exercise risk, provide exercise guidance, real-time monitoring, and feedback to the patient; The feedback optimization closed loop is used to continuously collect exercise execution data, physiological responses and periodic assessment results, and feed them back to the AI model to continuously iterate and optimize the prescription generation strategy for the patient. The data acquisition layer, intelligent analysis and decision-making layer, interactive execution layer, and feedback optimization closed loop are interconnected.
2. The rehabilitation exercise auxiliary decision-making system for hemodialysis patients according to claim 1, characterized in that: The data acquisition layer includes a hemodialysis machine interface, wearable devices, biochemical and medical record interfaces, and electronic scales.
3. The rehabilitation exercise auxiliary decision-making system for hemodialysis patients according to claim 2, characterized in that: The hemodialysis machine interface connects to the hemodialysis machine and is used to acquire real-time data such as the patient's heart rate, blood pressure, blood oxygen saturation, and ultrafiltration volume / rate. The wearable device includes either a smart bracelet or a chest patch. The biochemistry and medical record interface connects to the hospital information system and the laboratory information system to acquire key laboratory indicators such as dry weight, hemoglobin, serum potassium, and troponin, as well as diagnostic and medication history. The electronic scale guides patients to regularly fill out fatigue scales, depression and anxiety scales, quality of life scales, and subjective motor experience scales via a tablet computer or mobile APP.
4. The rehabilitation exercise auxiliary decision-making system for hemodialysis patients according to claim 1, characterized in that: The intelligent analysis and decision-making layer includes a patient digital profiling module, a hybrid risk assessment model, and a prescription generation model.
5. The rehabilitation exercise auxiliary decision-making system for hemodialysis patients according to claim 4, characterized in that: The patient profiling module is used to integrate all static and dynamic data to construct a dynamic digital profile for each patient, including underlying diseases, fitness level, risk labels, and exercise history. The hybrid risk assessment model adopts a hybrid model that combines rules and data-driven approaches. The rule base of the hybrid risk assessment model includes exercise contraindication rules based at least on real-time blood pressure changes, serum potassium concentration, and ultrafiltration rate. The prescription generation model uses a reinforcement learning algorithm, using periodic patient functional assessment results as reward signals to dynamically adjust exercise prescription parameters.
6. The rehabilitation exercise auxiliary decision-making system for hemodialysis patients according to claim 1, characterized in that: The interaction and execution layer includes a nurse console and a patient app. The nurse console is used to display an overview of the movement risks of all patients in the ward, individualized exercise prescription suggestions, and real-time alarm information. The patient app is used for patient exercise guidance, real-time monitoring and feedback, motivation and education.
7. A method for generating personalized exercise prescriptions for hemodialysis patients, characterized in that, The steps include: S1: Acquire and integrate multi-source patient data; S2: Run the hybrid risk assessment model; S3: If the risk is acceptable, call the AI prescription model to generate a recommended plan; S4: Output and execute through the human-computer interface in the human-computer interaction module.
8. The method for generating personalized exercise prescriptions for hemodialysis patients according to claim 7, characterized in that: The human-computer interaction module enables real-time two-way communication of physiological data and automatic intervention logic during the patient's movement.