A smart assessment system for individualized rehabilitation training after cardiac surgery

The intelligent assessment system for individualized rehabilitation training after cardiac surgery solves the problem of the lack of a unified basis for comparison of training records in the rehabilitation assessment after cardiac surgery. It enables the correlation analysis of changes in training load and changes in physiological response, thereby improving the accuracy of rehabilitation training assessment and the pertinence of parameter adjustment.

CN123120355APending Publication Date: 2026-11-17THE SECOND AFFILIATED HOSPITAL ARMY MEDICAL UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611351572.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-02
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

The current postoperative cardiac rehabilitation assessment lacks a unified comparative basis, and the correspondence between changes in training load and changes in physiological response is unclear, leading to interpretive biases due to different rehabilitation processes, making it difficult to accurately determine the training process and adjust training parameters.

Method used

Design an intelligent assessment system for individualized rehabilitation training after cardiac surgery, including a rehabilitation data acquisition module, a recovery stage identification module, a stage calibration processing module, a training tolerance trend analysis module, and an assessment result generation module. By acquiring and processing continuous training records, the system identifies recovery stages, performs stage alignment and consistency analysis, and generates individualized rehabilitation training assessment results.

Benefits of technology

Establish a basis for comparable analysis in continuous rehabilitation training, accurately identify training tolerance trends, improve the consistency between assessment results and patients' actual rehabilitation status, and enhance the pertinence and safety of training parameter adjustments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123120355A_ABST
    Figure CN123120355A_ABST
Patent Text Reader

Abstract

This invention relates to the field of postoperative rehabilitation assessment technology, specifically to an intelligent assessment system for individualized rehabilitation training after cardiac surgery. In this invention, a rehabilitation data acquisition module acquires rehabilitation-related status data of patients during rehabilitation training after cardiac surgery. A recovery stage identification module extracts recovery stage identifiers and determines the patient's current postoperative recovery stage. A stage calibration processing module then performs stage alignment, consistency analysis, and credibility assessment based on the current postoperative recovery stage, generating a stage calibration dataset. Subsequently, a training tolerance trend analysis module analyzes the stage coupling relationship between changes in training load and changes in physiological response, identifying the patient's training tolerance trend in the current postoperative recovery stage. Finally, an assessment result generation module distinguishes between adaptive fluctuations and abnormal tolerance changes, generating individualized rehabilitation training assessment results. This helps improve the comparability and relevance of continuous rehabilitation training assessments.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of postoperative rehabilitation assessment technology, specifically to an intelligent assessment system for individualized rehabilitation training after cardiac surgery. Background Technology

[0002] Post-cardiac surgery rehabilitation is a process of continuous adjustment of the training program as the body's functions gradually recover. During this process, the patient's training load tolerance, physiological response patterns, and post-training recovery performance typically change continuously with the rehabilitation progress, exhibiting strong temporal and phased characteristics. For the same patient, there is often a dynamic correlation between training duration, training distance, heart rate changes, recovery speed, fatigue sensations, and discomfort records at different time points. This necessitates that rehabilitation assessments not be based solely on single training data but require a comprehensive analysis of the changing patterns observed in continuous training records.

[0003] However, current postoperative rehabilitation assessments often lack a unified basis for comparison between training records before and after the procedure, and the correspondence between changes in training load and physiological responses is not clear enough. This leads to interpretative biases even when adjacent training results come from the same patient, due to differences in rehabilitation progress. Particularly when training ability gradually improves but fluctuates in the short term, changes in physiological responses after certain increases in load may be normal adaptations during rehabilitation or reflect a decline in tolerance. Without an assessment mechanism that combines continuous training status, the direction of response changes, and the evolution of recovery, it is difficult to accurately determine the training process and hinders the rational adjustment of subsequent training parameters.

[0004] Therefore, it is necessary to propose an intelligent assessment scheme for continuous rehabilitation scenarios after cardiac surgery, so that training records can be analyzed uniformly on a comparable basis, and further combine the correlation changes between load changes and physiological responses to make a more targeted assessment of the patient's rehabilitation training status. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent assessment system for individualized rehabilitation training after cardiac surgery to solve the problems mentioned in the background art. The specific technical problems include how to establish a comparable analysis basis for training data with time-series changes during continuous rehabilitation training after cardiac surgery, and to identify tolerance trends by combining the correlation between changes in training load and changes in physiological response, so as to solve the problem of difficulty in accurately distinguishing between adaptive fluctuations and abnormal tolerance changes.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent assessment system for individualized rehabilitation training after cardiac surgery, comprising a rehabilitation data acquisition module, a recovery stage identification module, a stage calibration processing module, a training tolerance trend analysis module, and an assessment result generation module, wherein: The rehabilitation data acquisition module is used to acquire rehabilitation-related status data of patients after cardiac surgery during rehabilitation training, specifically including: Acquire single training duration, single training distance, pre-training heart rate, peak heart rate during training, recovery heart rate at the first preset moment after training, fatigue score, chest tightness record, and training timestamp; The rehabilitation-related status data corresponding to the same training session are merged according to the training timestamp to form the current training record; Multiple current training records arranged in chronological order are formed into a continuous training record sequence; and rehabilitation-related status data are preprocessed, including time alignment, outlier screening, outlier data completion, and low-confidence record marking.

[0007] The rehabilitation data acquisition module is used to collect and organize rehabilitation-related status data during the continuous rehabilitation training process of patients after cardiac surgery. It acquires data such as single training duration, single training distance, pre-training heart rate, peak heart rate during training, recovery heart rate at the first preset time after training, fatigue score, chest tightness record, and training timestamp. It then merges the rehabilitation-related status data corresponding to the same training session according to the training timestamp to form the current training record. Multiple current training records arranged in chronological order are then formed into a continuous training record sequence. Combined with preprocessing methods such as time alignment, outlier screening, outlier data completion, and low-confidence record marking, discrete training information is gradually transformed into a continuous and temporally sequential analyzable data foundation. This provides a more unified data input condition for subsequent comparison, calibration, and trend identification of the relationship between changes in training load and changes in physiological response.

[0008] The recovery phase identification module is used to extract recovery phase identification features based on rehabilitation-related status data, specifically including: Heart rate response value per unit time is extracted based on single training duration, pre-training heart rate, and peak heart rate during training. Extract the heart rate response value per unit distance based on the single training distance, pre-training heart rate, and peak heart rate during training. Heart rate recovery rate is extracted based on peak heart rate during training, recovery heart rate at the first preset moment after training, and fixed recovery time. The fatigue trend slope is extracted based on the fatigue score in the preset continuous training observation window; and the frequency of chest tightness and the response fluctuation value are extracted based on the chest tightness record and the heart rate response value per unit time in the preset continuous training observation window.

[0009] The recovery stage identification module determines the patient's current postoperative recovery stage based on recovery stage identifier features, specifically including: The average heart rate response value per unit time, the average heart rate response value per unit distance, the average heart rate recovery rate, the fatigue trend slope, the frequency of chest tightness, and the response fluctuation value corresponding to the observation window are compared with the corresponding thresholds to obtain multiple stage scores. The comprehensive stage score is obtained by summing the scores of multiple stages; and the patient's current postoperative recovery stage is determined by matching the comprehensive stage score with the thresholds for the early postoperative recovery stage, the transitional postoperative recovery stage, the functional improvement stage, and the stable postoperative rehabilitation stage.

[0010] The recovery phase identification module is used to extract recovery phase identifier features based on rehabilitation-related state data, and to determine the patient's current postoperative recovery phase based on these features. This module extracts heart rate response values ​​per unit time, heart rate response values ​​per unit distance, heart rate recovery rate, fatigue trend slope, frequency of chest tightness, and response fluctuation values. It summarizes information reflecting the training load response pattern, state fluctuation direction, and recovery evolution state in the continuous training record sequence. Then, through multiple phase scores, comprehensive phase scores, and matching relationships with different postoperative recovery phase thresholds, it determines the patient's current recovery interval. This allows subsequent comparisons of training records to move beyond simple temporal relationships and instead be processed within an analytical context corresponding to the current postoperative recovery phase, thus improving the comparability between continuous training data.

[0011] The stage calibration processing module is used to perform stage alignment and consistency analysis on rehabilitation-related status data based on the determined current postoperative recovery stage. The stage alignment process specifically includes: The first candidate set is obtained by filtering historical training records from the historical training database that have the same current postoperative recovery stage as the current training record. The second candidate set is obtained by selecting historical training records from the first candidate set whose training duration difference from the current training record does not exceed the preset duration tolerance range and whose training distance difference from the current training record does not exceed the preset distance tolerance range. Calculate the relative deviations between the current training record and each historical training record in the second candidate set in terms of heart rate response value per unit time, heart rate response value per unit distance, heart rate recovery rate, and fatigue score, and obtain the mean alignment deviation based on the relative deviations; The historical training record with the smallest average alignment deviation is selected as the effective alignment object, and the current training record and the effective alignment object are used together to form the stage alignment sample.

[0012] The consistency analysis process in the stage calibration module specifically includes: Calculate the average values ​​of effective alignment objects in terms of training distance, heart rate response value per unit time, heart rate recovery rate, and fatigue score; Compare the current training record with the average value to determine the direction of change in training distance, the direction of change in heart rate response value per unit time, the direction of change in heart rate recovery rate, and the direction of change in fatigue score. When the training distance increases, and the change in heart rate response value per unit time does not exceed the preset load response tolerance threshold, and the heart rate recovery rate does not decrease, and the fatigue score does not increase, the first consistency marker is determined. The second consistency marker is determined when the training distance does not increase, the heart rate response value per unit time increases, the heart rate recovery rate decreases, and the fatigue score increases. Consistency scores are obtained based on the first and second consistency tags.

[0013] The phase calibration processing module performs credibility assessments based on information response relationships and change persistence relationships, generating a phase calibration dataset, which specifically includes: Information response relationship score is obtained based on consistency score; the direction of change of heart rate response value, heart rate recovery rate and fatigue score per unit time is judged based on preset short-term continuous observation window to obtain change continuity relationship score. The mean alignment deviation is subjected to a credibility mapping process to obtain an alignment credibility score; a comprehensive credibility score is obtained based on the information response relationship score, the change persistence relationship score, and the alignment credibility score. Based on the comprehensive credibility score and the preset credibility screening threshold, the current training records are included, marked as records to be reviewed, or marked as low-credibility records to generate a stage calibration dataset.

[0014] The stage calibration processing module is used to perform stage alignment and consistency analysis on rehabilitation-related status data based on the determined current postoperative recovery stage, and to conduct credibility assessment based on information response relationship and change persistence relationship, generating a stage calibration dataset. Firstly, it selects historical training records from the historical training database that have the same current postoperative recovery stage and whose training duration and distance are within a comparable range to the current training record, identifying effective alignment objects and forming stage alignment samples. This allows the current training record to obtain a reference basis under similar recovery states and training conditions. Secondly, it compares the direction of change in training distance, heart rate response value per unit time, heart rate recovery rate, and fatigue score, combining the information response relationship score, change persistence relationship score, and alignment credibility score to form a comprehensive credibility score. Records are then classified as included, marked as pending review, or marked as low-credibility records. This establishes a more targeted basis for comparable analysis during continuous rehabilitation training, reducing judgment bias caused by differences in training conditions or short-term abnormal records.

[0015] The training tolerance trend analysis module is used to analyze the stage coupling relationship between changes in training load and changes in physiological response in the stage calibration dataset based on recovery stage identifier features. Specifically, it includes: The training distance sequence, heart rate response value sequence, heart rate recovery rate sequence, and fatigue score sequence are constructed based on the most recent consecutive training records. The training distance sequence and the heart rate response value sequence per unit time, the training distance sequence and the fatigue score sequence, and the training distance sequence and the heart rate recovery rate sequence were standardized respectively to calculate the first coupling coefficient, the second coupling coefficient, and the third coupling coefficient. Based on the relationship between the first coupling coefficient, the second coupling coefficient, and the third coupling coefficient and the first stable coupling interval, the first fatigue coupling threshold, the first recovery coupling threshold, the second load response coupling threshold, the second fatigue coupling threshold, and the second recovery coupling threshold, the stage coupling state is determined to be stable coupling, abnormal coupling, or fluctuating coupling.

[0016] The training tolerance trend analysis module identifies the patient's training tolerance trend during the current postoperative recovery phase, specifically including: Based on the preset multi-window trend analysis unit, the trend identification is performed on the stage coupling state, the average heart rate response value per unit time, the average heart rate recovery rate, and the average fatigue score in multiple continuous analysis windows. Based on the changes in stage coupling state in multiple continuous analysis windows and the slope of changes corresponding to the average heart rate response value, average heart rate recovery rate and average fatigue score per unit time, the training tolerance trend can be identified as gradually improving, basically stable, continuously declining or adapting to fluctuations.

[0017] The training tolerance trend analysis module is used to analyze the stage coupling relationship between changes in training load and changes in physiological response in the stage calibration dataset based on recovery stage identifier features, and to identify the training tolerance trend of patients in the current postoperative recovery stage. This training tolerance trend analysis module constructs training distance sequences, unit time heart rate response value sequences, heart rate recovery rate sequences, and fatigue score sequences, and calculates the first coupling coefficient, second coupling coefficient, and third coupling coefficient, respectively, to determine whether the stage coupling state is stable coupling, abnormal coupling, or fluctuating coupling. Then, it combines the changes in stage coupling state in multiple continuous analysis windows with the slope of the changes corresponding to the average unit time heart rate response value, average heart rate recovery rate, and average fatigue score to identify the training tolerance trend. This allows the correlation between changes in training load and changes in physiological response to be comprehensively expressed within a continuous time series, thereby providing a trend-level analytical basis for distinguishing between short-term adaptive fluctuations and persistent abnormal tolerance changes.

[0018] The assessment results generation module is used to distinguish between adaptive fluctuations and abnormal tolerance changes based on training tolerance trends, and to generate individualized rehabilitation training assessment results, specifically including: Receive training tolerance trend, overall credibility score, deviation of current training record from the mean of effective alignment object per unit time heart rate response value, heart rate recovery rate deviation, fatigue score deviation, and frequency of chest tightness occurrence; The degree of abnormality is obtained based on the deviation of heart rate response value per unit time, the deviation of heart rate recovery rate, and the deviation of fatigue score; Anomalies were classified based on training tolerance trends, abnormality scores, frequency of chest tightness, and overall reliability scores to distinguish between adaptive fluctuations and abnormal tolerance changes. Based on the current postoperative recovery stage and training tolerance trend, individualized rehabilitation training assessment results and training parameter adjustment results are generated.

[0019] The assessment result generation module is used to distinguish between adaptive fluctuations and abnormal tolerance changes based on training tolerance trends, and to generate individualized rehabilitation training assessment results. This module receives training tolerance trends, comprehensive reliability scores, deviations of the current training record from the mean of the effective alignment object in terms of unit-time heart rate response value, heart rate recovery rate deviation, fatigue score deviation, and frequency of chest tightness. It first obtains an abnormality value based on the deviations in unit-time heart rate response value, heart rate recovery rate deviation, and fatigue score deviation, and then performs abnormality classification by combining the training tolerance trend, abnormality value, frequency of chest tightness, and comprehensive reliability score. This allows for a more targeted judgment on the patient's current training performance under the current postoperative recovery stage, and outputs individualized rehabilitation training assessment results and training parameter adjustment results, so as to make a more detailed distinction between normal adaptive changes and abnormal tolerance changes that need attention during continuous rehabilitation training.

[0020] Compared with the prior art, the beneficial effects of the present invention are: This invention provides an intelligent assessment system for individualized rehabilitation training after cardiac surgery. Through the coordinated operation of a rehabilitation data acquisition module, a recovery stage identification module, a stage calibration processing module, a training tolerance trend analysis module, and an assessment result generation module, it can first determine the patient's current postoperative recovery stage by combining recovery stage identification features with rehabilitation-related state data with temporal changes during continuous rehabilitation training after cardiac surgery. Then, based on the current postoperative recovery stage, it performs stage alignment, consistency analysis, and credibility assessment on rehabilitation-related state data to generate a stage calibration dataset. Furthermore, it analyzes the stage coupling relationship between changes in training load and changes in physiological response to identify the patient's training tolerance trend in the current postoperative recovery stage. This establishes a more reasonable basis for comparable analysis of training records in continuous training scenarios, improves the differentiation and targeting of adaptive fluctuations and abnormal tolerance changes, and makes the individualized rehabilitation training assessment results and training parameter adjustment results better fit the patient's rehabilitation process. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the core process of the overall module of the present invention; Figure 2 This is a schematic diagram of the core process for extracting identification features during the recovery phase of this invention; Figure 3 This is a schematic diagram of the core process of applying the recovery phase identification feature in this invention; Figure 4 This is a comparative analysis chart of the effects of the present invention.

[0022] In the diagram: 100, Rehabilitation Data Acquisition Module; 200, Recovery Stage Identification Module; 300, Stage Calibration Processing Module; 400, Training Tolerance Trend Analysis Module; 500, Assessment Result Generation Module. Detailed Implementation

[0023] The technical solutions in 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.

[0024] This invention provides an intelligent assessment system for individualized rehabilitation training after cardiac surgery. This system dynamically analyzes changes in training load, physiological response, recovery ability, and symptoms in patients undergoing rehabilitation training. It then identifies the patient's current postoperative recovery stage label and generates individualized rehabilitation training assessment results and training parameter adjustment results that match the current postoperative recovery stage label. The system is deployed on a server, doctor's workstation, rehabilitation training terminal, mobile terminal, cloud platform, or in a computing environment formed by a combination of the above devices.

[0025] Please see Figure 1 A smart assessment system for individualized rehabilitation training after cardiac surgery includes a rehabilitation data acquisition module 100, a recovery stage identification module 200, a stage calibration processing module 300, a training tolerance trend analysis module 400, and an assessment result generation module 500. Among them: The rehabilitation data acquisition module 100 is used to acquire rehabilitation-related status data of patients after cardiac surgery during rehabilitation training; the recovery stage identification module 200 is used to extract recovery stage identification features based on rehabilitation-related status data and determine the patient's current postoperative recovery stage label based on the recovery stage identification features; the stage calibration processing module 300 is used to perform stage alignment and consistency analysis on rehabilitation-related status data based on the current postoperative recovery stage label, and to perform credibility assessment based on information response relationship and change persistence relationship, generating a stage calibration dataset; the training tolerance trend analysis module 400 is used to analyze the stage coupling relationship between training load changes and physiological response changes in the stage calibration dataset based on recovery stage identification features, and to identify the patient's training tolerance trend under the current postoperative recovery stage; the evaluation result generation module 500 is used to distinguish between adaptive fluctuations and abnormal tolerance changes based on the training tolerance trend, and to generate individualized rehabilitation training evaluation results and training parameter adjustment results.

[0026] Through the modular setup described above, this application can transform the assessment of rehabilitation training for post-cardiac surgery patients from a static judgment based on post-operative time to a dynamic judgment based on continuous training load, physiological response, recovery ability, and symptom evolution. This improves the consistency between the assessment results and the patient's actual rehabilitation status, and enhances the pertinence and safety of training parameter adjustment results.

[0027] The rehabilitation data acquisition module 100 is used to acquire rehabilitation-related status data of post-cardiac surgery patients during rehabilitation training; the rehabilitation-related status data includes single training session duration, single training session distance, pre-training heart rate, peak heart rate during training, recovery heart rate at the first preset time after training, fatigue score, chest tightness record, and training timestamp; among which: The duration of a single training session is used to characterize the duration of a single training session; the distance of a single training session is used to characterize the external load level of a single training session; the pre-training heart rate is used to characterize the baseline physiological state before training begins; the peak heart rate during training is used to characterize the maximum physiological load response during training; the recovery heart rate at the first preset time after training is used to characterize the recovery status after training; the fatigue score is used to characterize the patient's subjective tolerance to the training load; the chest tightness record is used to characterize whether a symptom event occurred; and the training timestamp is used to establish the temporal sequence relationship between consecutive training records.

[0028] In this embodiment, the rehabilitation data acquisition module 100 acquires rehabilitation-related status data from wearable monitoring devices, rehabilitation training devices, mobile terminal input interfaces, hospital information systems, and electronic medical record systems to build a multi-source training monitoring data foundation covering training load, immediate physiological response, recovery status, and symptom events.

[0029] After acquisition, the rehabilitation data acquisition module 100 merges the data corresponding to the same training session according to the training timestamp to form the current training record; and forms a continuous training record sequence by arranging multiple current training records in chronological order.

[0030] The rehabilitation data acquisition module 100 is also used to preprocess the acquired rehabilitation-related status data. The preprocessing process includes time alignment, outlier screening, outlier data imputation, and low-confidence record marking, as detailed below: Time-aligned all rehabilitation-related status data according to training timestamps; Screen for outliers in pre-training heart rate, peak heart rate during training, and recovery heart rate at the first preset time after training. When there is only one abnormal data in the current training record, the data is filled in by the average of the corresponding fields in the previous and next valid training records. If there are two or more outlier data points in the current training record, the current training record will be marked as a low-confidence record.

[0031] After preprocessing, the rehabilitation data acquisition module 100 outputs a continuous training record sequence for use by subsequent modules.

[0032] Please see Figure 2The recovery stage identification module 200 is used to extract recovery stage identification features based on rehabilitation-related state data, and to determine the patient's current postoperative recovery stage label based on the recovery stage identification features. The recovery stage identification features characterize the patient's load response pattern, state fluctuation direction, and recovery evolution status under the current postoperative recovery stage. These features include heart rate response value per unit time, heart rate response value per unit distance, heart rate recovery rate, fatigue trend slope, frequency of chest tightness, and response fluctuation value. Using the duration of a single training session, the pre-training heart rate, and the peak heart rate during training as inputs, the heart rate increment is first calculated. ,Right now ,in Indicates the peak training heart rate. This represents the heart rate before training; further, the heart rate response value per unit time is calculated. ,Right now ,in This represents the duration of a single training session and the heart rate response value per unit time. Used to characterize the intensity of physiological response corresponding to a unit of training time.

[0033] Using the single training distance, pre-training heart rate, and peak heart rate during training as inputs, calculate the heart rate response value per unit distance. ,Right now ,in This represents the distance of a single training session; heart rate response value per unit distance. Used to characterize the physiological response intensity corresponding to a unit training distance.

[0034] Using the peak heart rate during training, the recovery heart rate at the first preset moment after training, and a fixed recovery time as inputs, the heart rate recovery amount is first calculated. ,Right now ,in This represents the recovery heart rate at the first preset time after training; further, the heart rate recovery rate is calculated. ,Right now ,in This indicates a fixed recovery time; in this embodiment, the fixed recovery time is five minutes, and the heart rate recovery rate is... Used to characterize the recovery speed after training ends.

[0035] Using the fatigue scores and corresponding training numbers from the five most recent training records as input, sequential values ​​(first, second, third, fourth, and fifth) are assigned in chronological order. Each current training record's fatigue score is then mapped one-to-one with its sequential value, forming five sets of data for trend calculation. The sequential values ​​represent training progress, and the corresponding fatigue scores represent fatigue changes. A least-squares linear fitting method is used to determine the straight line that best reflects the overall trend of the five sets of data, and the slope of this line is used as the fatigue trend slope. When the fatigue trend slope A value greater than zero indicates that the fatigue score is generally trending upward; when the slope of the fatigue trend is... When the value is less than zero, it indicates that the fatigue score is generally declining.

[0036] The recovery phase identification module 200 extracts the fatigue trend slope, frequency of chest tightness occurrence, and response fluctuation value based on a preset continuous training observation window. In this embodiment, the preset continuous training observation window includes the five most recent current training records. Using the five most recent current training records as the observation window can balance the sensitivity to short-term changes and the stability of continuous samples. The frequency of chest tightness occurrence is obtained by statistically analyzing the chest tightness records in the observation window. The standard deviation of the heart rate response value per unit time in the observation window is calculated to obtain the response fluctuation value. It is used to characterize the stability of the training response.

[0037] After obtaining the characteristics identifying the recovery phase, the average heart rate response value per unit time, the average heart rate response value per unit distance, the average heart rate recovery rate, the fatigue trend slope, the frequency of chest tightness, and the response fluctuation value corresponding to the observation window are compared with the corresponding thresholds to obtain six phase scores. The six phase scores are then summed to obtain a comprehensive phase score. Comprehensive stage score This is used to map multidimensional training loads and physiological response parameters to a unified stage representation, so that subsequent modules can perform historical sample screening and stage calibration based on the unified stage label.

[0038] Based on the comprehensive stage score The study established a correlation between training load, physiological response, recovery ability, and symptom changes at different postoperative recovery stages. Thresholds for the early postoperative recovery stage, the transitional postoperative recovery stage, the functional improvement stage, and the stable postoperative rehabilitation stage were pre-set, and a comprehensive stage score was calculated. The current postoperative recovery stage label is determined by matching the threshold values ​​of each stage; where: The threshold value of the early postoperative recovery stage corresponds to the comprehensive stage score. Scores in the 5-6 range correspond to the overall stage score of the postoperative transitional recovery phase. Scores in the 3-4 range correspond to the overall stage score of the postoperative functional improvement stage. Scores in the 1-2 range correspond to the comprehensive stage score of the postoperative stable recovery stage threshold. The score is between 0 and 1. When the comprehensive stage score When the early postoperative recovery stage threshold is met, an early postoperative recovery stage label is generated; when the overall stage score is... When the postoperative transitional recovery stage threshold is met, a postoperative transitional recovery stage label is generated; when the overall stage score is... When the postoperative functional improvement stage threshold is met, a postoperative functional improvement stage label is generated; when the overall stage score is... When the postoperative stable recovery stage threshold is met, a postoperative stable recovery stage label is generated. Output recovery stage identifier features and current postoperative recovery stage label.

[0039] Please see Figure 3 The stage calibration processing module 300 is used to perform stage alignment and consistency analysis on rehabilitation-related status data based on the current postoperative recovery stage label, and to perform credibility assessment based on information response relationship and change persistence relationship to generate stage calibration dataset. The stage calibration processing is used to perform same-stage constraints and credibility screening on continuous training monitoring data to reduce the entry of abnormally collected data, single occasional fluctuation data and stage mismatch data into the subsequent trend analysis process.

[0040] Specifically, the stage calibration processing module 300 receives the current postoperative recovery stage label, recovery stage identification features, and current training records output by the recovery stage identification module 200, and sequentially performs stage alignment, consistency analysis, information response relationship score calculation, change persistence relationship score calculation, comprehensive credibility score calculation, and stage calibration dataset generation; the stage alignment process is as follows: First, historical training records with the same current postoperative recovery stage label as the current training record are selected from the historical training database to obtain the first candidate set; Secondly, from the first candidate set, further filter historical training records where the difference between the training duration and the current training record does not exceed the preset duration tolerance range, and the difference between the training distance and the current training record does not exceed the preset distance tolerance range, to obtain the second candidate set; Then, calculate the relative deviations between the current training record and each historical training record in the second candidate set in terms of heart rate response value per unit time, heart rate response value per unit distance, heart rate recovery rate, and fatigue score; for any historical training record, the first... Relative deviation of the indicators The calculation formula is: ,in, Indicates the number of times in the current training record Item index value, This indicates the first in the historical training record. Item indicator value.

[0041] The mean alignment deviation for this historical training record is obtained by averaging the relative deviations of the four indicators. ,Right now ,in , , and These represent the relative deviations of the four indicators.

[0042] The stage calibration processing module 300 selects the historical training record with the smallest average alignment deviation as the effective alignment object, and combines the current training record with the effective alignment object to form the stage alignment sample.

[0043] The consistency analysis process is as follows: After completing the stage alignment, a consistency analysis is performed on the stage aligned samples, specifically including: First, calculate the average values ​​of the training distance, heart rate response value per unit time, heart rate recovery rate, and fatigue score of the first three valid training records. Then, compare the current training record with the above average values ​​to determine the direction of change in training distance, heart rate response value per unit time, heart rate recovery rate, and fatigue score.

[0044] When the training distance increases, and the change in heart rate response value per unit time does not exceed the preset load response tolerance threshold, and the heart rate recovery rate does not decrease, and the fatigue score does not increase, a first consistency marker is recorded. Otherwise, record the first consistency tag. The purpose of setting a preset load response tolerance threshold is to allow for small, normal physiological fluctuations when the training load increases.

[0045] A second consistency marker is recorded when the training distance does not increase, the heart rate response value per unit time increases, the heart rate recovery rate decreases, and the fatigue score increases. Otherwise, record the second consistency tag. ; Based on this, the consistency score is calculated. ,Right now ,in The value can be 0, 1, or 2, and the consistency score is... Used to quantify the consistency between changes in current training load and changes in physiological response.

[0046] The stage calibration processing module 300 calculates the information response relationship score based on the consistency score. The information response relationship score is used to characterize the degree of matching between changes in the current training load and changes in the physiological response. The specific rules are as follows: when At that time, the information response relationship will be scored. Recorded as 1; when At that time, the information response relationship will be scored. Recorded as 0.5; when At that time, the information response relationship will be scored. Recorded as 0; Therefore, information response relationship scoring Information response relationship score is in the range of 0 to 1. Used as input parameters for calculating the overall credibility score.

[0047] The stage calibration processing module 300, based on a preset short-term continuous observation window, continuously judges the direction of change of heart rate response value, heart rate recovery rate, and fatigue score per unit time to obtain a score of the continuity of change. Specifically, it includes: In this embodiment, the short-term continuous observation window includes the three most recent current training records; When the heart rate response value per unit time continuously increases within the short-term sustained observation window, the first sustained sub-score is recorded. Otherwise, remember ; When the heart rate recovery rate continuously decreases within the short-term continuous observation window, the second continuity sub-score is recorded. Otherwise, remember ; When the fatigue score continuously increases within the short-term continuous observation window, the third continuous sub-score is recorded. Otherwise, remember .

[0048] Persistence of Change Score The calculation formula is: Therefore, the change persistence relationship score The score for the duration of change is in the range of 0 to 1. It is used to characterize whether anomalous changes are continuous.

[0049] After obtaining the information response relationship score Persistence of Change Score and the mean of alignment deviation Subsequently, the stage calibration processing module 300 further calculates the comprehensive credibility score.

[0050] To ensure that the mean alignment deviation is included in the unified dimensional calculation, the mean alignment deviation is first subjected to a reliable mapping process to obtain the alignment reliability score. Alignment confidence score The calculation formula is: , among which when At that time, Recorded as 0; when At that time, It is denoted as 1 to ensure the alignment reliability score. It is in the range of 0 to 1.

[0051] Subsequently, the stage calibration processing module 300 calculates the overall credibility score using a weighted summation method. Overall credibility score The calculation formula is: Where 0.5 is the weight for the information response relationship score, 0.3 is the weight for the alignment credibility score, and 0.2 is the weight for the change persistence relationship score; In this embodiment, the above weight configuration is adopted because the information response relationship score directly reflects the degree of matching between the current training load change and the physiological response change, and can more directly reflect whether the current training record conforms to the normal response pattern under the current postoperative recovery stage, so it is given the highest weight; the alignment credibility score is used to reflect the consistency between the current training record and the historical training record under the same current postoperative recovery stage label, and can provide an important reference for the credibility of the current training record, so it is given the second highest weight; the change persistence relationship score is mainly used to reflect whether the abnormal change persists in continuous training, and its direct discrimination effect on the current single training record is relatively weak, so it is given a relatively low weight.

[0052] In other embodiments, the above weights can also be adjusted based on the statistical results of historical training samples, the results of clinical experience calibration, the identification requirements of different postoperative recovery stages, or different rehabilitation training scenarios.

[0053] After obtaining the overall credibility score Subsequently, the stage calibration processing module 300 filters the current training records based on the comprehensive credibility score and the preset credibility screening threshold. The preset credibility screening threshold includes a credibility inclusion threshold and a credibility verification threshold, and the screening rules are as follows: when If the current training record is not less than the credible inclusion threshold, it will be included in the stage calibration dataset. when When the value falls between the credible verification threshold and the credible inclusion threshold, the current training record is marked as a record to be verified. when If the value is less than the confidence verification threshold, the current training record will be marked as a low-confidence record and will not be included in the stage calibration dataset. The stage calibration dataset consists of current training records that meet the inclusion criteria based on the comprehensive confidence score, and is used for subsequent training tolerance trend analysis.

[0054] The stage calibration dataset consists of current training records whose comprehensive confidence scores meet the inclusion criteria. It is used by the subsequent training tolerance trend analysis module 400 to perform stage coupling relationship analysis and training tolerance trend identification.

[0055] The training tolerance trend analysis module 400 is used to analyze the stage coupling relationship between changes in training load and changes in physiological response in the stage calibration dataset based on recovery stage identification features, and to identify the training tolerance trend of patients in the current postoperative recovery stage.

[0056] Specifically, the training tolerance trend analysis module 400 receives the phase calibration dataset and the recovery phase identification features output by the recovery phase identification module 200, and constructs a training distance sequence, a heart rate response value sequence per unit time, a heart rate recovery rate sequence, and a fatigue score sequence based on the five most recent current training records.

[0057] The training tolerance trend analysis module 400 first standardizes the training distance sequence and the heart rate response value sequence per unit time to calculate the first coupling coefficient; then it standardizes the training distance sequence and the fatigue score sequence to calculate the second coupling coefficient; and finally it standardizes the training distance sequence and the heart rate recovery rate sequence to calculate the third coupling coefficient.

[0058] After obtaining the first coupling coefficient, the second coupling coefficient, and the third coupling coefficient, the training tolerance trend analysis module 400 outputs the stage coupling state according to the preset judgment rules, specifically including: When the first coupling coefficient is in the first stable coupling range, the second coupling coefficient is less than the first fatigue coupling threshold, and the third coupling coefficient is greater than the first recovery coupling threshold, the stage coupling state is determined to be stable coupling. When the first coupling coefficient is not less than the second load response coupling threshold, the second coupling coefficient is not less than the second fatigue coupling threshold, and the third coupling coefficient is less than the second recovery coupling threshold, the stage coupling state is determined to be abnormal coupling. In other cases, the stage coupling state is determined to be fluctuating coupling.

[0059] Among them, the first fatigue coupling threshold is less than the second fatigue coupling threshold, the first recovery coupling threshold is greater than the second recovery coupling threshold, the first stable coupling interval is used to characterize the normal coupling range between training load changes and physiological response changes, and the second load response coupling threshold is used to characterize the abnormally enhanced coupling range between training load changes and physiological response changes.

[0060] Furthermore, the training tolerance trend analysis module 400, based on a preset multi-window trend analysis unit, performs trend identification on the stage coupling state and the average value of key indicators in multiple continuous analysis windows. In this embodiment, the preset window trend analysis unit preferably includes the three most recent analysis windows, each including five consecutive current training records. Specifically, using the stage coupling state, average heart rate response value per unit time, average heart rate recovery rate, and average fatigue score of the three most recent analysis windows as inputs, the corresponding slope of change is calculated respectively, and the training tolerance trend is identified accordingly. Specifically, this includes: When the phase coupling states of the three most recent analysis windows are fluctuation coupling, stable coupling, and stable coupling in sequence, and the slope of the average change of the heart rate response value per unit time is less than zero, the slope of the average change of the heart rate recovery rate is greater than zero, and the slope of the average change of the fatigue score is less than zero, the training tolerance trend is determined to be gradually improving. When the stage coupling states of the three most recent analysis windows are all stable couplings, and the absolute values ​​of the slopes of the three changes do not exceed the preset stable trend threshold, the training tolerance trend is determined to be basically stable. When at least two of the three most recent analysis windows are in anomalous coupling, and the slope of the average change of the heart rate response value per unit time is greater than zero, the slope of the average change of the heart rate recovery rate is less than zero, and the slope of the average change of the fatigue score is greater than zero, the training tolerance trend is determined to be a continuous decline. In other cases, the training tolerance trend is determined to be adaptive fluctuation; Training tolerance trends are used to reflect the direction of a patient's adaptation to training load and the degree of change in tolerance during the current postoperative recovery phase.

[0061] The assessment result generation module 500 is used to distinguish between adaptive fluctuations and abnormal tolerance changes based on training tolerance trends, and to generate individualized rehabilitation training assessment results and training parameter adjustment results.

[0062] Specifically, the evaluation result generation module 500 receives the training tolerance trend output by the training tolerance trend analysis module 400, the comprehensive credibility score output by the stage calibration processing module 300, and the deviation of the current training record from the mean of the effective alignment object in terms of unit time heart rate response value, heart rate recovery rate deviation, fatigue score deviation, and the frequency of chest tightness in the last three current training records.

[0063] The evaluation result generation module 500 first calculates the anomaly level value based on preset weights. Abnormality value The calculation formula is: ,in, This indicates the deviation of the heart rate response value per unit time. This indicates a deviation in heart rate recovery rate. This indicates the fatigue score deviation; where 0.4 is the weight of the deviation in heart rate response value per unit time, 0.3 is the weight of the deviation in heart rate recovery rate, and 0.3 is the weight of the fatigue score deviation. In this embodiment, the above weighting configuration is adopted because the deviation of the heart rate response value per unit time can more directly reflect the patient's immediate physiological stress changes under the current training load and has a high sensitivity to the identification of abnormal tolerance changes, so it is given a high weight; the deviation of the heart rate recovery rate is used to reflect the changes in recovery ability after training, which can reflect whether the patient's recovery reserve has decreased; the deviation of the fatigue score is used to reflect the patient's subjective tolerance to the training load, which can help identify tolerance abnormalities when physiological abnormalities have not yet fully manifested; since changes in recovery ability and subjective fatigue feelings are both of important reference value for the identification of abnormal tolerance changes, but are slightly less direct than the immediate physiological response, they are given the same and slightly lower weights.

[0064] In other embodiments, the above weights can be adjusted based on the statistical results of historical rehabilitation training samples, clinical experience calibration results, risk stratification results of different patients, or the contribution of each indicator to the identification of abnormal tolerance changes under different training modes.

[0065] Subsequently, the evaluation result generation module 500 performs anomaly classification based on training tolerance trends, anomaly severity values, frequency of chest tightness occurrences, and comprehensive reliability scores to obtain the judgment results, specifically including: When the training tolerance trend is gradually improving, the frequency of chest tightness is zero times, and the abnormality value is less than the abnormality judgment threshold, the judgment result is stable improvement; When the training tolerance trend is basically stable, the abnormality value is less than the abnormality judgment threshold, and the frequency of chest tightness is no more than once, the judgment result is adaptive fluctuation. When the training tolerance trend is continuously decreasing, the abnormality value is not less than the abnormality judgment threshold, the frequency of chest tightness is not less than twice, and the comprehensive credibility score is not less than the credibility confirmation threshold, the judgment result is abnormal tolerance change. When the training tolerance trend is adaptive fluctuation, the abnormality value is not less than the abnormality judgment threshold, and there are two chest tightness records in the last three current training records, the judgment result is also abnormal tolerance change. The judgment results are used to distinguish between normal fluctuation states that can continue to adapt and abnormal tolerance states that require training restrictions or review interventions.

[0066] After obtaining the judgment result, the evaluation result generation module 500 further combines the current postoperative recovery stage label and training tolerance trend to generate individualized rehabilitation training evaluation results and training parameter adjustment results. In this embodiment, the training parameter adjustment results specifically include: When the current postoperative recovery stage is labeled as the postoperative functional improvement stage, the training tolerance trend is gradually improving, and the judgment result is stable improvement, the training parameter adjustment result is output. The output training parameter adjustment result is to increase the single training duration by three minutes and increase the single training distance by one hundred meters. When the current postoperative recovery stage is labeled as the postoperative transitional recovery stage, the training tolerance trend is basically stable, and the judgment result is adaptive fluctuation, the training parameter adjustment result is output. The output training parameter adjustment result is to maintain the current single training duration and the current single training distance, and to continuously observe the current training record three times. When the current postoperative recovery stage is labeled as the early postoperative recovery stage, the training tolerance trend is continuously decreasing, and the judgment result is abnormal tolerance change, the training parameter adjustment result is output. The output training parameter adjustment result is to reduce the single training time by five minutes and reduce the single training distance by one hundred meters, and prompt for clinical review. The training parameter adjustment results are used to convert the evaluation results into executable training control parameters to improve the individual fit and safety of training load adjustment.

[0067] In this embodiment, the thresholds in the above process include: early postoperative recovery stage threshold, postoperative transitional recovery stage threshold, postoperative functional improvement stage threshold, postoperative stable rehabilitation stage threshold, preset duration tolerance range, preset distance tolerance range, preset load response tolerance threshold, credible inclusion threshold, credible verification threshold, first stable coupling interval, first fatigue coupling threshold, first recovery coupling threshold, second load response coupling threshold, second fatigue coupling threshold, second recovery coupling threshold, preset stable trend threshold, anomaly determination threshold, and credible confirmation threshold; wherein: The threshold value of the early postoperative recovery stage corresponds to the comprehensive stage score. Scores in the 5-6 range correspond to the overall stage score of the postoperative transitional recovery phase. Scores in the 3-4 range correspond to the overall stage score of the postoperative functional improvement stage. Scores in the 1-2 range correspond to the comprehensive stage score of the postoperative stable recovery stage threshold. Scores in the range of 0 to 1 or below; The preset duration tolerance range is preferably 1 minute to 3 minutes, and the preset distance tolerance range is preferably 50 meters to 150 meters; A preset load response tolerance threshold is used to limit the allowable range of changes in heart rate response value per unit time when training distance increases, preferably 5% to 15% of the corresponding baseline heart rate response value per unit time; The credible inclusion threshold is preferably between 0.65 and 0.85, and the credible verification threshold is preferably between 0.30 and 0.50, with the credible verification threshold being less than the credible inclusion threshold; The first stable coupling range is preferably 0.20 to 0.60, the first fatigue coupling threshold is preferably 0.20 to 0.40, the first recovery coupling threshold is preferably 0.10 to 0.30, the second load response coupling threshold is preferably 0.55 to 0.75, the second fatigue coupling threshold is preferably 0.45 to 0.65, and the second recovery coupling threshold is preferably -0.10 to 0.10, wherein the first fatigue coupling threshold is less than the second fatigue coupling threshold and the first recovery coupling threshold is greater than the second recovery coupling threshold; A preset stable trend threshold is used to limit the allowable fluctuation range of the slope of the average change of the heart rate response value per unit time, the slope of the average change of the heart rate recovery rate, and the slope of the average change of the fatigue score. It is preferably 0.01 to 0.10 of the absolute value of each corresponding slope. An anomaly detection threshold is used to distinguish between adaptive fluctuations and abnormal tolerance changes, and is preferably 0.40 to 0.70; the credible confirmation threshold is preferably 0.60 to 0.85. The determination of the above thresholds is preferably based on the statistical results of historical rehabilitation training samples, the results of clinical experience calibration, the changing patterns between training load and physiological response at different postoperative recovery stages, the results of patient risk stratification, and the data distribution characteristics under different rehabilitation training scenarios. Furthermore, it can be dynamically corrected by combining the data acquisition accuracy of wearable monitoring devices, the load grading accuracy of rehabilitation training devices, and the fluctuation characteristics of continuous training record sequences, so that each threshold has both a data statistical basis and a clinical application basis.

[0068] Please see Figure 4 To verify the effectiveness of this embodiment, this embodiment was compared with a static judgment method based on postoperative time and a rehabilitation assessment method based on a single training physiological threshold under the same test set. The same test set consisted of test set 1 to test set 5, and each method used the same data source, the same sample division method, and the same result comparison standard to ensure the objectivity and comparability of the comparison results.

[0069] Under the same test set, using the consistency between the assessment results and the patient's actual recovery status as the comparison index, the results of this embodiment on test sets 1 to 5 were 90.7%, 88.9%, 92.6%, 89.8%, and 91.5%, respectively, with an average of 90.7%. The results of the static judgment method based on postoperative time on test sets 1 to 5 were 77.8%, 80.6%, 78.4%, 81.3%, and 79.1%, respectively, with an average of 79.4%. The results of the rehabilitation assessment method based on the physiological threshold of a single training session on test sets 1 to 5 were 84.3%, 82.5%, 85.7%, 83.6%, and 84.8%, respectively, with an average of 84.2%.

[0070] Meanwhile, this embodiment outperforms static judgment methods based on postoperative time and rehabilitation assessment methods based on single training physiological thresholds in terms of consistency between assessment results and patients' actual recovery status, the pertinence and safety of training parameter adjustment results, and the accuracy of distinguishing between adaptive fluctuations and abnormal tolerance changes. The reason is that this embodiment establishes a continuous analysis relationship around recovery stage identification features, current postoperative recovery stage labels, stage calibration datasets, and training tolerance trends, which can perform collaborative analysis on dynamic load changes, physiological response changes, and symptom evolution in patients' rehabilitation training. Therefore, the obtained assessment results are more consistent with the patients' actual recovery status.

[0071] Should Figure 4 The figure shows a comparison of the consistency between the assessment results and the patient's actual recovery status in this embodiment, the static judgment method based on postoperative time, and the rehabilitation assessment method based on the physiological threshold of a single training session, under the same test set. As can be seen from the figure, this embodiment shows high consistency under different test sets, and although there are normal fluctuations caused by changes in the test set, it is still significantly better than the two existing methods overall, thus indicating that this embodiment has a better dynamic assessment effect.

[0072] The execution process of this embodiment is illustrated below with a simplified example, specifically including: A patient who underwent heart surgery completed five consecutive rehabilitation training sessions. The rehabilitation data acquisition module 100 acquired the single training duration, single training distance, pre-training heart rate, peak heart rate during training, recovery heart rate at the first preset time after training, fatigue score, chest tightness record, and training timestamp for each training session. After completing time alignment and outlier screening, a continuous training record sequence was formed. The recovery phase identification module 200 calculates the heart rate response value per unit time, heart rate response value per unit distance, heart rate recovery rate, fatigue trend slope, frequency of chest tightness, and response fluctuation value based on the continuous training record sequence. It is assumed that the six phase scores are accumulated to obtain a comprehensive phase score. If the value is 2, then the current postoperative recovery stage label is determined as the postoperative functional improvement stage based on the postoperative functional improvement stage threshold. Subsequently, the stage calibration processing module 300 selects historical training records from the historical training database that are consistent with the current postoperative recovery stage label and have similar training duration and training distance as effective alignment objects, and calculates the average alignment deviation between the current training record and the effective alignment objects. A comprehensive credibility score was obtained by combining the consistency analysis results and the change persistence relationship score. ,like If the credible inclusion threshold is reached, the current training record will be included in the stage calibration dataset. Next, the training tolerance trend analysis module 400 calculates the first coupling coefficient, the second coupling coefficient, and the third coupling coefficient based on the training distance sequence, the heart rate response value sequence, the heart rate recovery rate sequence, and the fatigue score sequence formed by the most recent five current training records. If the first coupling coefficient is in the first stable coupling interval, the second coupling coefficient is less than the first fatigue coupling threshold, and the third coupling coefficient is greater than the first recovery coupling threshold, then the stage coupling state is determined to be stable coupling. In the most recent three analysis windows, it further identifies that the slope of the average change of the heart rate response value per unit time is less than zero, the slope of the average change of the heart rate recovery rate is greater than zero, and the slope of the average change of the fatigue score is less than zero, thereby determining that the training tolerance trend is gradually improving. Finally, the evaluation result generation module 500 combines the anomaly degree value. The frequency of chest tightness and the overall reliability score are used to classify abnormalities. If the frequency of chest tightness is zero and the abnormality value is less than the abnormality judgment threshold, the judgment result is stable improvement. Under the condition that the current postoperative recovery stage is labeled as the postoperative functional improvement stage, the training parameter adjustment result is output, that is, the single training time is increased by three minutes and the single training distance is increased by one hundred meters. In this way, the complete execution process is completed from the acquisition of multi-source rehabilitation-related state data, identification of the current postoperative recovery stage, stage calibration, training tolerance trend analysis to the generation of individualized rehabilitation training evaluation results and training parameter adjustment results.

[0073] As can be seen from the above description, the intelligent assessment system for individualized rehabilitation training after cardiac surgery provided in this embodiment has the following technical effects: The rehabilitation data acquisition module 100 organizes and preprocesses rehabilitation-related status data during continuous rehabilitation training of post-cardiac surgery patients. The recovery stage identification module 200 uses recovery stage identifier features to determine the current post-operative recovery stage label. The stage calibration processing module 300 then performs stage alignment, consistency analysis, and reliability assessment on the rehabilitation-related status data around the current post-operative recovery stage label, generating a stage calibration dataset. Based on this, the training tolerance trend analysis module 400 further analyzes the stage coupling relationship between changes in training load and changes in physiological response in the stage calibration dataset based on recovery stage identifier features, identifying the patient's training tolerance trend in the current post-operative recovery stage. Finally, the evaluation results generate... Module 500 combines training tolerance trends, comprehensive reliability scores, and deviation and symptom information to distinguish between adaptive fluctuations and abnormal tolerance changes, and generates individualized rehabilitation training assessment results and training parameter adjustment results. Therefore, the system forms two interrelated technical means around the characteristics of the recovery stage identification: determining the current postoperative recovery stage label and identifying the training tolerance trend. This enables continuous training data to establish a comparable analysis relationship based on matching the current postoperative recovery stage label, and improves the specificity of distinguishing between adaptive fluctuations and abnormal tolerance changes by combining the correlation between changes in training load and changes in physiological response. As a result, the individualized rehabilitation training assessment results and training parameter adjustment results are more in line with the dynamic changes in the patient's rehabilitation training process.

[0074] 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 present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent assessment system for individualized rehabilitation training after cardiac surgery, characterized in that, It includes a rehabilitation data acquisition module, a recovery phase identification module, a phase calibration processing module, a training tolerance trend analysis module, and an assessment result generation module, among which: The rehabilitation data acquisition module is used to acquire rehabilitation-related status data of patients after cardiac surgery during rehabilitation training. The recovery stage identification module is used to extract recovery stage identification features based on the rehabilitation-related status data, and determine the patient's current postoperative recovery stage based on the recovery stage identification features; The stage calibration processing module is used to perform stage alignment and consistency analysis on the rehabilitation-related status data based on the determined current postoperative recovery stage, and to perform credibility assessment based on information response relationship and change persistence relationship, and generate a stage calibration dataset. The training tolerance trend analysis module is used to analyze the stage coupling relationship between changes in training load and changes in physiological response in the stage calibration dataset based on the recovery stage identification features, and to identify the training tolerance trend of the patient in the current postoperative recovery stage. The assessment result generation module is used to distinguish between adaptive fluctuations and abnormal tolerance changes based on the training tolerance trend, and to generate individualized rehabilitation training assessment results.

2. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 1, characterized in that, The process of acquiring the rehabilitation-related status data specifically includes: Acquire single training duration, single training distance, pre-training heart rate, peak heart rate during training, recovery heart rate at the first preset moment after training, fatigue score, chest tightness record, and training timestamp; According to the training timestamp, the various rehabilitation-related status data corresponding to the same training session are merged to form the current training record; Multiple current training records arranged in chronological order are formed into a continuous training record sequence; and the rehabilitation-related status data are preprocessed, including time alignment, outlier screening, outlier data completion, and low-confidence record marking.

3. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 2, characterized in that, The extraction process of the identification features in the recovery phase specifically includes: Heart rate response value per unit time is extracted based on single training duration, pre-training heart rate, and peak heart rate during training. Extract the heart rate response value per unit distance based on the single training distance, pre-training heart rate, and peak heart rate during training. Heart rate recovery rate is extracted based on peak heart rate during training, recovery heart rate at the first preset moment after training, and fixed recovery time. The fatigue trend slope is extracted based on the fatigue score in the preset continuous training observation window; and the frequency of chest tightness and the response fluctuation value are extracted based on the chest tightness record and the heart rate response value per unit time in the preset continuous training observation window, respectively.

4. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 3, characterized in that, The process of determining the current postoperative recovery stage specifically includes: The average heart rate response value per unit time, the average heart rate response value per unit distance, the average heart rate recovery rate, the fatigue trend slope, the frequency of chest tightness, and the response fluctuation value corresponding to the observation window are compared with the corresponding thresholds to obtain multiple stage scores. The multiple stage scores are summed to obtain a comprehensive stage score; and the patient's current postoperative recovery stage is determined based on the matching relationship between the comprehensive stage score and the thresholds for the early postoperative recovery stage, the transitional postoperative recovery stage, the functional improvement stage, and the stable postoperative rehabilitation stage.

5. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 1, characterized in that, The stage alignment process specifically includes: The first candidate set is obtained by filtering historical training records from the historical training database that have the same current postoperative recovery stage as the current training record. From the first candidate set, select historical training records where the difference between the training duration and the current training record does not exceed a preset duration tolerance range and the difference between the training distance and the current training record does not exceed a preset distance tolerance range to obtain a second candidate set; Calculate the relative deviations between the current training record and each historical training record in the second candidate set in terms of heart rate response value per unit time, heart rate response value per unit distance, heart rate recovery rate, and fatigue score, and obtain the mean alignment deviation based on the relative deviations; The historical training record with the smallest average alignment deviation is selected as the effective alignment object, and the current training record and the effective alignment object are used together to form the stage alignment sample.

6. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 5, characterized in that, The consistency analysis process specifically includes: Calculate the average values ​​of effective alignment objects in terms of training distance, heart rate response value per unit time, heart rate recovery rate, and fatigue score; The current training record is compared with the average value to determine the direction of change in training distance, the direction of change in heart rate response value per unit time, the direction of change in heart rate recovery rate, and the direction of change in fatigue score. When the training distance increases, and the change in heart rate response value per unit time does not exceed the preset load response tolerance threshold, and the heart rate recovery rate does not decrease, and the fatigue score does not increase, the first consistency marker is determined. The second consistency marker is determined when the training distance does not increase, the heart rate response value per unit time increases, the heart rate recovery rate decreases, and the fatigue score increases. A consistency score is obtained based on the first consistency tag and the second consistency tag.

7. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 6, characterized in that, The process of generating the stage calibration dataset specifically includes: The information response relationship score is obtained based on the consistency score; the change direction of heart rate response value, heart rate recovery rate and fatigue score per unit time is judged based on the preset short-term continuous observation window to obtain the change continuity relationship score. The mean alignment deviation is subjected to a credibility mapping process to obtain an alignment credibility score; a comprehensive credibility score is obtained based on the information response relationship score, the change persistence relationship score, and the alignment credibility score. Based on the comprehensive credibility score and the preset credibility screening threshold, the current training records are included, marked as records to be reviewed, or marked as low-credibility records to generate a stage calibration dataset.

8. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 1, characterized in that, The process by which the training tolerance trend analysis module analyzes the stage coupling relationship specifically includes: The training distance sequence, heart rate response value sequence, heart rate recovery rate sequence, and fatigue score sequence are constructed based on the most recent consecutive training records. The training distance sequence and the heart rate response value sequence per unit time, the training distance sequence and the fatigue score sequence, and the training distance sequence and the heart rate recovery rate sequence were standardized respectively to calculate the first coupling coefficient, the second coupling coefficient, and the third coupling coefficient. Based on the relationship between the first coupling coefficient, the second coupling coefficient, and the third coupling coefficient and the first stable coupling interval, the first fatigue coupling threshold, the first recovery coupling threshold, the second load response coupling threshold, the second fatigue coupling threshold, and the second recovery coupling threshold, the stage coupling state is determined to be stable coupling, abnormal coupling, or fluctuating coupling.

9. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 8, characterized in that, The process of identifying the training tolerance trend specifically includes: Based on the preset multi-window trend analysis unit, the trend identification is performed on the stage coupling state, the average heart rate response value per unit time, the average heart rate recovery rate, and the average fatigue score in multiple continuous analysis windows. Based on the changes in stage coupling state in multiple continuous analysis windows and the slope of change corresponding to the average heart rate response value per unit time, the average heart rate recovery rate, and the average fatigue score, the training tolerance trend is identified as gradually increasing, basically stable, continuously decreasing, or adapting to fluctuations.

10. The intelligent assessment system for individualized rehabilitation training after cardiac surgery according to claim 1, characterized in that, The process of generating the individualized rehabilitation training assessment results specifically includes: Receive training tolerance trend, overall credibility score, deviation of current training record from the mean of effective alignment object per unit time heart rate response value, heart rate recovery rate deviation, fatigue score deviation, and frequency of chest tightness occurrence; An abnormality value is obtained based on the deviation of the heart rate response value per unit time, the deviation of the heart rate recovery rate, and the deviation of the fatigue score; Anomaly classification is performed based on the training tolerance trend, the abnormality value, the frequency of chest tightness, and the comprehensive reliability score to distinguish between adaptive fluctuations and abnormal tolerance changes. Based on the current postoperative recovery stage and training tolerance trend, individualized rehabilitation training assessment results and training parameter adjustment results are generated.