A postoperative respiratory function intelligent evaluation and rehabilitation guidance system for thoracic surgery

Abstract

This invention discloses an intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery, belonging to the field of intelligent medical rehabilitation systems. The invention collects postoperative time-dimensional data, recovery progress data, and real-time respiratory data from patients in real time through a data acquisition module. Based on this data, a dynamic prediction model module is used to construct a prediction model that can predict the trend of changes in the patient's respiratory function 3-7 days in advance. This ability to predict in advance allows medical staff to intervene promptly, adjust rehabilitation plans, and effectively prevent the occurrence of potential respiratory dysfunction, thereby promoting rapid patient recovery. For example, the system can identify the risk of alveolar hypoventilation or airway spasm in advance and automatically generate a proactive rehabilitation adjustment plan to ensure that patients receive timely and accurate treatment.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical rehabilitation systems, specifically to an intelligent assessment and rehabilitation guidance system for respiratory function after thoracic surgery. Background Technology

[0002] With the continuous advancement of medical technology, thoracic surgery has become an important means of treating a variety of chest diseases. However, the recovery of respiratory function after surgery is directly related to the patient's quality of recovery and quality of life. Since thoracic surgery often involves the operation of important organs in the thoracic cavity, postoperative patients often face risks such as limited respiratory function, insufficient alveolar ventilation, and airway spasm.

[0003] Traditional postoperative respiratory function assessment techniques in thoracic surgery have several shortcomings: First, the assessment cycle is lengthy, typically relying on regular outpatient follow-ups or daily monitoring during hospitalization. This makes it difficult for medical staff to promptly grasp the dynamic changes in the patient's respiratory function, potentially missing the optimal intervention window. Second, the assessment methods are highly subjective, relying heavily on the experience and judgment of medical staff, lacking objective and quantitative assessment standards, leading to significant individual differences in assessment results. Third, traditional techniques struggle to provide personalized rehabilitation guidance. Due to the lack of comprehensive analysis of individual patient characteristics, real-time physiological data, and subjective feedback, the developed rehabilitation plans often lack specificity and fail to meet the specific needs of different patients. Finally, traditional assessment methods have limitations in preventing potential respiratory dysfunction, failing to identify and intervene in possible respiratory problems in advance, thus affecting the overall rehabilitation outcome for patients.

[0004] Given the shortcomings of traditional postoperative respiratory function assessment techniques in thoracic surgery, it is therefore of great importance to develop an intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery. By integrating a data acquisition module, a dynamic prediction model module, a risk assessment module, and a rehabilitation plan generation module, this system can achieve a comprehensive and accurate assessment of the patient's postoperative respiratory function. This system can not only predict the trend of changes in the patient's respiratory function 3-7 days in advance, providing medical staff with timely intervention basis, but also generate personalized rehabilitation adjustment plans by combining the patient's individual characteristics, subjective feedback, and real-time physiological data, effectively improving the patient's rehabilitation experience and overall effect.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery, the system comprising the following components: a data acquisition module, a dynamic prediction model module, a risk assessment module, and a rehabilitation plan generation module; The data acquisition module collects postoperative time data, recovery progress data, and real-time respiratory data from patients. It also includes wearable device acquisition units and electronic medical record acquisition units, and performs filtering, format standardization, and encrypted transmission processing on the collected data. The dynamic prediction model module is constructed based on the collected data and uses a time-weighted respiratory function prediction algorithm to build a prediction model that can predict the trend of changes in the patient's respiratory function 3-7 days in advance and supports iterative optimization of model parameters. The risk assessment module determines whether a patient is at risk of alveolar hypoventilation or airway spasm, and can make the assessment by linking the patient’s preoperative respiratory reserve data and real-time physiological fluctuation data. The rehabilitation plan generation module automatically generates a preliminary rehabilitation adjustment plan when a risk is detected, and personalizes it by combining the patient's individual characteristics and subjective feedback.

[0007] Furthermore, the data acquisition module is implemented in the following ways: the wearable device acquisition unit uses a flexible piezoelectric film sensor to acquire the patient's respiratory electromyography (EMG) signals at a frequency of 200Hz. The acquired raw signals are first filtered by a 50Hz notch filter to remove power frequency interference, and then filtered by a 0.5-30Hz bandpass filter to extract the effective EMG signals. The electronic medical record acquisition unit connects to the hospital's HIS system and rehabilitation management system to automatically extract postoperative time stamp data, wound healing level in the recovery progress, average daily respiratory training completion rate, and postoperative complication history. All acquired data is processed to unify the format, converted to a 16-bit numerical format, and then transmitted to the dynamic prediction model module through an encrypted transmission protocol. The transmission delay is controlled within 50ms to avoid data lag affecting the prediction results.

[0008] Furthermore, the specific implementation steps for the wearable device linkage requirement of the data acquisition module are as follows: the wearable device and the system use the Bluetooth Low Energy 5.1 protocol for data transmission to ensure a data transmission latency of <100ms, while using edge-side AES-256 encryption to protect the patient's privacy data; after the system generates a rehabilitation plan, the wearable device will adjust its monitoring frequency according to the content of the plan; the triaxial accelerometer built into the wearable device will collect the patient's action data during rehabilitation training, and identify whether the breathing training actions are standard through a preset action feature library. If non-standard actions are identified, voice correction instructions will be pushed in real time; the action data will be synchronously transmitted to the system's risk assessment module as supplementary data for risk assessment to optimize the accuracy of risk assessment.

[0009] Furthermore, the time-weighted respiratory function prediction algorithm in the dynamic prediction model module is formulated as follows: ,in This is a predictive value for respiratory function, ranging from 0 to 100. A higher value indicates better respiratory function. The postoperative time dimension weights were obtained by fitting the respiratory function changes at different stages after surgery data of 1200 patients after thoracic surgery using a time-weighted particle swarm optimization algorithm. The values ​​ranged from 0.2 to 0.4, with higher weights for shorter postoperative times. The weighting of recovery progress was obtained by fitting the correlation data of the patient's wound healing level and the average daily respiratory training completion rate using the least squares method, with a value range of 0.3-0.5. The better the recovery progress, the lower the weight. The weights for real-time respiratory data are calculated from the variance of respiratory electromyography signals and the coefficient of variation of respiratory flow rate, with a value range of 0.2-0.3. The greater the fluctuation in real-time respiratory data, the higher the weight. The values ​​are quantified based on the postoperative time dimension: 1 for 1-3 days postoperatively, 2 for 4-7 days postoperatively, 3 for 8-14 days postoperatively, and 4 for more than 14 days postoperatively. To quantify the recovery progress, the wound healing level and the average daily breathing training completion rate are converted into a value of 0-4. The values ​​are quantified from real-time respiratory data, converted from the mean respiratory electromyography signal and the mean respiratory flow rate to a value of 0-4. The time-related correction factor is calculated from the ratio of the number of days after surgery to the preoperative respiratory baseline, and its value ranges from 0.8 to 1.2.

[0010] Furthermore, the iterative optimization of model parameters in the dynamic prediction model module employs an interpolation-driven weight optimization algorithm, with the following formula: ,in The optimized postoperative time dimension weighting, The weights before optimization; The difference rate between the actual and predicted values ​​of the patient's respiratory function within this iteration period is: Difference rate = (Actual respiratory function value - Predicted respiratory function value) / Predicted respiratory function value. To optimize the coefficients, they are determined by the improvement in prediction accuracy on the test set, with a value ranging from 0.05 to 0.2. hour, Take 0.05, when hour, Set the value to 0.2; the model iteration module has an iteration cycle of 7 days. During this cycle, the actual respiratory function data and rehabilitation plan execution data of patients are collected. After outlier removal, the data are input into the formula for weight optimization. The optimized model will be validated in a test set of 200 patients who did not participate in the initial training. If the prediction accuracy improves by more than 5%, the parameters of the dynamic prediction model will be updated.

[0011] Furthermore, the risk assessment module is implemented by first taking the respiratory function prediction values ​​output by the dynamic prediction model module. Compared with the preset postoperative respiratory function benchmark values ​​for patients of the same surgical type and age group. For comparison, F0 was determined by the 95% confidence interval of the mean postoperative respiratory function in 1000 similar patients; if The system further extracts features associated with alveolar hypoventilation and airway spasm from the prediction results, and determines the peak fluctuation of respiratory electromyography (EMG) signals >20μV. Risk scores are then calculated for each feature: alveolar hypoventilation risk score = (F0-F) / F0×100, and airway spasm risk score = (peak fluctuation of respiratory EMG signals -20μV) / 20μV×100. When either risk score exceeds a preset threshold, the risk assessment module generates a risk correlation report, identifies the triggering factors of the risk, and transmits it synchronously to the rehabilitation plan generation module.

[0012] Furthermore, the preoperative data association requirement of the risk assessment module also includes a preoperative respiratory reserve data association assessment step, specifically: the preoperative respiratory reserve data is calculated from the vital capacity and maximum voluntary ventilation of the preoperative pulmonary function test, and the calculation method is preoperative respiratory reserve value = (preoperative vital capacity + preoperative maximum voluntary ventilation) / 2; the risk assessment module first compares the preoperative respiratory reserve value with the postoperative respiratory function prediction value F. If F is lower than 60% of the preoperative respiratory reserve value, the risk assessment threshold for alveolar hypoventilation and airway spasm is lowered by 10%, that is, the risk threshold for alveolar hypoventilation is adjusted from 60% to 54%, and the risk threshold for airway spasm is adjusted from 50% to 45%; then, combined with the fluctuation amplitude of the real-time respiratory electromyography signal, if the fluctuation amplitude is >30μV, the risk warning level is further increased from warning to high warning; the final risk result will mark the risk-related factors, such as preoperative respiratory reserve insufficiency + large postoperative respiratory electromyography fluctuation, which facilitates targeted intervention by medical staff.

[0013] Furthermore, the rehabilitation plan generation module is implemented in the following ways: First, the rehabilitation plan generation module receives the risk association report from the risk assessment module. Combining the patient's postoperative time dimension data and recovery progress data, it determines the core direction of rehabilitation adjustments. If the risk is alveolar hypoventilation and the patient is in the 1-7 day postoperative stage, the frequency of airway cleaning training is adjusted from twice a day to four times a day. If the patient is in the 8 days or more postoperative stage, the rhythm of breathing training is adjusted from every 20 seconds to every 30 seconds. If the risk is airway spasm, breathing assistance techniques are promoted first, while increasing the rest interval of breathing training. After the plan is generated, it is first pushed to the management terminal of medical staff, with a 12-hour review window. If the medical staff does not make any modifications, it is automatically pushed to the patient's wearable device, and a check-in reminder for plan execution is generated every 6 hours.

[0014] Furthermore, the personalized adaptation requirements of the rehabilitation program generation module also include personalized adaptation steps based on the patient's age and body mass index (BMI). Specifically: if the patient is older than 65 years and has a BMI greater than 28, the intensity of rehabilitation training will be reduced by 20%, while the rest interval for breathing training will be increased from 2 minutes of rest after every 10 minutes of training to 3 minutes of rest after every 8 minutes of training, to avoid respiratory muscle fatigue or wound traction pain. If the patient is younger than 40 years and has a BMI less than 18, the intensity of rehabilitation training will be increased by 10%, while the number of sets of breathing training will be increased from 3 sets per day to 4 sets per day, to improve the recovery speed of respiratory function. In addition, based on the patient's dietary data, if the patient's protein intake is insufficient, guidance on respiratory muscle nutrition supplementation will be added to the rehabilitation program, such as increasing the intake of eggs, milk, and lean meat. All adapted and adjusted programs will generate corresponding execution record tables, recording the reasons for the program adjustments, the content of the adjustments, and the patient's execution status, for subsequent model iterations.

[0015] Furthermore, the system also includes a patient feedback module. The specific implementation steps are as follows: the patient provides feedback on their feelings during the rehabilitation training process through voice input on the wearable device or the system's official mini-program. The system converts the patient's feedback data into a quantitative feedback score, ranging from 0 to 10, with higher scores indicating greater discomfort. This score is then simultaneously transmitted to the rehabilitation plan generation module. If the feedback score is greater than 7, the rehabilitation plan generation module will immediately adjust the intensity and duration of the rehabilitation training, reducing the training duration by 30% and simultaneously pushing soothing breathing and relaxation techniques, such as slow pursed-lip breathing.

[0016] Compared with existing technologies, this intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery has the following beneficial effects: I. This system collects postoperative time-dimensional data, recovery progress data, and real-time respiratory data from patients in real time through the data acquisition module. Based on this data, the dynamic prediction model module builds a prediction model that can predict the trend of changes in patients' respiratory function 3-7 days in advance. This ability to predict in advance allows medical staff to intervene in a timely manner, adjust rehabilitation plans, and effectively prevent the occurrence of potential respiratory dysfunction, thereby promoting rapid patient recovery. For example, the system can identify the risk of alveolar hypoventilation or airway spasm in advance and automatically generate a pre-rehabilitation adjustment plan to ensure that patients receive timely and accurate treatment.

[0017] Second, this system not only possesses powerful predictive and intervention capabilities, but also achieves personalized adaptation of rehabilitation plans through a rehabilitation plan generation module. The system combines the patient's individual characteristics, subjective feedback, and real-time physiological data to tailor a rehabilitation training plan for each patient. For example, for older patients with a high BMI, the system will appropriately reduce the intensity of rehabilitation training and increase rest intervals to avoid respiratory muscle fatigue; while for younger patients with a lower BMI, the system may increase the training intensity to accelerate respiratory function recovery. Furthermore, the system allows patients to provide feedback on their training experience through wearable devices or a mini-program, and the system will immediately adjust the rehabilitation plan based on this feedback, thereby improving the patient's rehabilitation experience and overall effectiveness.

[0018] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0020] Figure 1 This is an overall flowchart of an intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery. Figure 2 A flowchart illustrating the data acquisition and wearable device linkage of an intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery. Figure 3 This is an architecture diagram of an intelligent assessment and rehabilitation guidance system for respiratory function after thoracic surgery. Detailed Implementation

[0021] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0022] Example 1 The overall system workflow is as follows: Figure 1 As shown, the collaborative architecture of the various modules of this system is as follows: Figure 3 As shown, firstly, the system's data acquisition module starts working: the wearable device, equipped with a flexible piezoelectric film sensor, acquires the patient's respiratory electromyography (EMG) signals at a sampling frequency of 200Hz. The acquired raw signals are first filtered by a 50Hz notch filter to remove power frequency interference, and then filtered by a 0.5-30Hz bandpass filter to extract the effective EMG signals. Simultaneously, the electronic medical record acquisition unit connects to the hospital's HIS system and rehabilitation management system, automatically extracting information such as the patient's postoperative timestamp data, wound healing level, daily average respiratory training completion rate, and history of no postoperative complications. After all acquired data undergoes standardized format processing, it is transmitted via Bluetooth Low Energy 5.1 protocol, with transmission latency controlled within 100ms. End-side AES-256 encryption is used to protect patient privacy data, and finally, the data is transmitted to the dynamic prediction model module. The data acquisition and wearable device linkage process is as follows... Figure 2 As shown.

[0023] After receiving the data, the dynamic prediction model module uses a time-weighted respiratory function prediction algorithm to predict the trend of changes in the patient's respiratory function, predicting the patient's respiratory function status 3-7 days in advance. The formula is as follows: ,in This is a predicted value for respiratory function; Postoperative time dimension weighting; To restore progress weight; Weighting of real-time respiratory data; Quantified values ​​for the postoperative time dimension; To restore the progress quantification value; Quantified values ​​for real-time respiratory data; For the time-related correction coefficient, the model has been optimized multiple times using an interpolation-driven weight optimization algorithm, with an iteration cycle of 7 days. The formula is as follows: ,in The optimized postoperative time dimension weighting, The weights before optimization; The percentage difference between the actual and predicted values ​​of the patient's respiratory function within this iteration cycle; To optimize the coefficients, the previously optimized model was validated on a test set of 200 patients who did not participate in the initial training, and the prediction accuracy improved by more than 5%. The parameters have been updated to ensure the accuracy of the current prediction results.

[0024] The risk assessment module processes the received respiratory function prediction values: First, it compares the predicted values ​​with preset postoperative respiratory function baseline values ​​for patients of the same surgical type and age group, finding that the predicted values ​​are lower than the baseline values. Then, it further extracts alveolar hypoventilation and airway spasm-related features from the prediction results, detecting a peak fluctuation of 28μV in the patient's respiratory electromyography (EMG) signal. After calculating the two risk scores, it is found that the alveolar hypoventilation risk score exceeds a preset threshold of 60%. Simultaneously, the risk assessment module correlates the patient's preoperative respiratory reserve data, finding that the postoperative respiratory function prediction value is lower than 60% of the preoperative respiratory reserve value. Therefore, the alveolar hypoventilation risk assessment threshold is lowered by 10%, from 60% to 54%. Combined with the real-time respiratory EMG signal fluctuation amplitude of 28μV not exceeding 30μV, the risk warning level is finally set to warning, a risk correlation report is generated, risk triggering factors are clearly marked, and the report is simultaneously transmitted to the rehabilitation plan generation module.

[0025] After receiving the risk-related report, the rehabilitation plan generation module, combined with the patient's postoperative time-dimensional data and recovery progress data, determined that the core direction of rehabilitation adjustments was to improve alveolar hypoventilation. Given that the patient was in the 1-7 day post-surgery stage, the frequency of airway cleaning training was prioritized to be increased from twice daily to four times daily. Simultaneously, considering the patient's age of over 65 years and BMI greater than 28, according to personalized adaptation rules, the intensity of rehabilitation training was reduced by 20%, and the rest interval for breathing training was adjusted from 2 minutes of rest after every 10 minutes of training to 3 minutes of rest after every 8 minutes of training. Dietary data from the system revealed insufficient protein intake, so guidance on respiratory muscle nutrition supplementation was added to the rehabilitation plan. Once the plan was generated, it was immediately pushed to the medical staff's management terminal, with a 12-hour review window. If the medical staff did not make any modifications within the review period, the plan was automatically pushed to the patient's wearable device, and a check-in reminder for plan execution was generated. During rehabilitation training, the wearable device's built-in triaxial accelerometer collects training movement data. Through a pre-set movement feature library, it identifies instances where abdominal breathing exercises are not performed correctly. The device then sends real-time voice correction commands and simultaneously transmits the movement data to the risk assessment module as supplementary data to optimize the accuracy of subsequent risk assessments. After training, the patient provides feedback on their experience through the wearable device's voice input function. The system converts this feedback into a quantified score of 8 points. The rehabilitation plan generation module immediately adjusts the intensity and duration of the rehabilitation training based on this feedback, further optimizing the rehabilitation plan.

[0026] Example 2 The overall operation of the system follows the following Figure 1 The process shown, and the collaborative architecture of each module are as follows: Figure 3As shown, the data acquisition module begins operation: the wearable device's flexible piezoelectric film sensor acquires the patient's respiratory electromyography (EMG) signals at a 200Hz acquisition frequency. The raw signals are filtered by a 50Hz notch filter to remove power frequency interference, and then filtered by a 0.5-30Hz bandpass filter to extract the effective EMG signals. The electronic medical record acquisition unit connects to the hospital's relevant systems, automatically extracting data such as the patient's postoperative timestamp, wound healing level, daily average respiratory training completion rate, and history of no postoperative complications. After the acquired data is formatted uniformly, it is encrypted and transmitted via Bluetooth Low Energy 5.1 protocol with a transmission delay controlled within 100ms, successfully transmitting it to the dynamic prediction model module. Details of the data acquisition and wearable device linkage are as follows... Figure 2 As shown.

[0027] The dynamic prediction model module uses a time-weighted respiratory function prediction algorithm to process the collected postoperative time-dimensional data, recovery progress data, and real-time respiratory data of patients to predict the trend of changes in patients' respiratory function. The model follows an optimization rule of one iteration cycle every 7 days. Previously, the model parameters were iteratively optimized using the difference-driven weight optimization algorithm. The optimized model was validated in a set of 200 independent test cases, and the prediction accuracy was improved by more than 5%. The parameters have been updated to ensure the reliability of the prediction results.

[0028] The risk assessment module compared the predicted respiratory function value with the postoperative respiratory function baseline value for patients of the same surgical type and age group, and found that the predicted value was lower than the baseline value. Further extraction of alveolar hypoventilation and airway spasm-related features revealed a peak fluctuation of 35μV in the patient's respiratory electromyography (EMG) signal. After calculating the risk score, the airway spasm risk score exceeded the preset threshold of 50%. Subsequently, by linking the patient's preoperative respiratory reserve data, it was found that the postoperative respiratory function predicted value was lower than 60% of the preoperative respiratory reserve value. Therefore, the airway spasm risk assessment threshold was lowered by 10% from 50% to 45%. Combined with the real-time respiratory EMG signal fluctuation amplitude exceeding 30μV (35μV), the risk warning level was raised from warning to high warning, a risk association report was generated, and the risk triggering factors were marked before being transmitted to the rehabilitation plan generation module.

[0029] After receiving the report, the rehabilitation plan generation module, combining the patient's postoperative 10-day time dimension data and recovery progress data, determined that the core direction of rehabilitation adjustment was to alleviate airway spasm. Since the patient was more than 8 days postoperatively and the risk was airway spasm, breathing assistance techniques were prioritized for the patient, and the rest intervals for breathing training were increased. Considering that the patient was under 40 years old and had a BMI of less than 18, the intensity of rehabilitation training was increased by 10% according to the personalized adaptation rules, and the number of breathing training sets was adjusted from 3 sets per day to 4 sets per day. Through dietary data verification, the patient's protein intake was sufficient, and no additional nutritional supplementation was recommended. After the plan was generated, it was pushed to the medical staff's management terminal. If the medical staff did not raise any modification opinions within the 12-hour review window, the plan was automatically pushed to the patient's wearable device, and a check-in reminder was generated at the same time.

[0030] During the patient's rehabilitation training, the wearable device's three-axis accelerometer collected motion data. This data was identified using a pre-set motion feature library, and all breathing training movements were deemed standard, with no voice correction commands pushed. The motion data was simultaneously transmitted to the risk assessment module as supplementary data. After training, the patient provided feedback on their experience through the system's official mini-program. The system converted this feedback into a quantitative score of 6 points, and the rehabilitation plan generation module did not immediately adjust the plan. The model then enters its next 7-day iteration cycle. During this cycle, the system collects the patient's actual respiratory function data and rehabilitation plan execution data. After removing outliers, the model weights are optimized using a difference-driven weight optimization algorithm. Once optimization is complete, the prediction accuracy will be validated on a 200-case test set. If the improvement exceeds 5%, the model parameters will be updated.

[0031] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A smart assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery, characterized in that, The system comprises the following components: a data acquisition module, a dynamic prediction model module, a risk assessment module, and a rehabilitation plan generation module; The data acquisition module collects postoperative time data, recovery progress data, and real-time respiratory data from patients. It also includes wearable device acquisition units and electronic medical record acquisition units, and performs filtering, format standardization, and encrypted transmission processing on the collected data. The dynamic prediction model module is constructed based on the collected data and uses a time-weighted respiratory function prediction algorithm to build a prediction model that can predict the trend of changes in the patient's respiratory function 3-7 days in advance and supports iterative optimization of model parameters. The risk assessment module determines whether a patient is at risk of alveolar hypoventilation or airway spasm, and can make the assessment by linking the patient’s preoperative respiratory reserve data and real-time physiological fluctuation data. The rehabilitation plan generation module automatically generates a preliminary rehabilitation adjustment plan when a risk is detected, and personalizes it by combining the patient's individual characteristics and subjective feedback.

2. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The data acquisition module is implemented as follows: the wearable device acquisition unit uses a flexible piezoelectric film sensor to acquire the patient's respiratory electromyography (EMG) signals at a frequency of 200Hz. The acquired raw signals are first filtered by a 50Hz notch filter to remove power frequency interference, and then filtered by a 0.5-30Hz bandpass filter to extract the effective EMG signals. The electronic medical record acquisition unit connects to the hospital's HIS system and rehabilitation management system to automatically extract postoperative time stamp data, wound healing level in the recovery progress, daily respiratory training completion rate, and postoperative complication history. All acquired data is processed in a standardized format and transmitted to the dynamic prediction model module via an encrypted transmission protocol.

3. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The specific implementation steps for the wearable device linkage requirement of the data acquisition module are as follows: the wearable device and the system use the Bluetooth Low Energy 5.1 protocol for data transmission to ensure that the data transmission latency is <100ms, and the patient's privacy data is protected by AES-256 encryption on the end side; after the system generates a rehabilitation plan, the wearable device will adjust its monitoring frequency according to the content of the plan; the triaxial accelerometer built into the wearable device will collect the patient's action data for performing rehabilitation training, and identify whether the breathing training actions are standard through a preset action feature library. If the action is not standard, a voice correction command will be pushed in real time. Action data will be synchronously transmitted to the system's risk assessment module as supplementary data for risk assessment, thereby optimizing the accuracy of risk assessment.

4. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The time-weighted respiratory function prediction algorithm in the dynamic prediction model module is formulated as follows: ,in This is a predicted value for respiratory function; Postoperative time dimension weighting; To restore progress weight; Weighting of real-time respiratory data; Quantified values ​​for the postoperative time dimension; To restore the progress quantification value; Quantified values ​​for real-time respiratory data; This is the time-related correction coefficient.

5. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The iterative optimization of model parameters in the dynamic prediction model module employs an interpolation-driven weight optimization algorithm, with the following formula: ,in The optimized postoperative time dimension weighting, The weights before optimization; The percentage difference between the actual and predicted values ​​of the patient's respiratory function within this iteration cycle; To optimize the coefficients, hour, Take 0.05, when hour, Set the value to 0.2; the model iteration module has an iteration cycle of 7 days. During this cycle, the actual respiratory function data and rehabilitation plan execution data of patients are collected. After outlier removal, the data are input into the formula for weight optimization. The optimized model will be validated in a test set of 200 patients who did not participate in the initial training. If the prediction accuracy improves by more than 5%, the parameters of the dynamic prediction model will be updated.

6. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The risk assessment module is implemented in the following steps: First, the respiratory function prediction value output by the dynamic prediction model module is... Compared with the preset postoperative respiratory function benchmark values ​​for patients of the same surgical type and age group. Compare; if The system further extracts features associated with alveolar hypoventilation and airway spasm from the prediction results, and determines the peak fluctuation of respiratory electromyography (EMG) signals >20μV. Risk scores are then calculated for each feature: alveolar hypoventilation risk score = (F0-F) / F0×100, and airway spasm risk score = (peak fluctuation of respiratory EMG signals -20μV) / 20μV×100. When either risk score exceeds a preset threshold, the risk assessment module generates a risk correlation report, identifies the triggering factors of the risk, and transmits it synchronously to the rehabilitation plan generation module.

7. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The preoperative data association requirement of the risk assessment module also includes a preoperative respiratory reserve data association assessment step, specifically: preoperative respiratory reserve data is calculated from vital capacity and maximum ventilation volume obtained from preoperative pulmonary function tests; the risk assessment module first compares the preoperative respiratory reserve value with the postoperative respiratory function prediction value F. If F is lower than 60% of the preoperative respiratory reserve value, the risk assessment threshold for alveolar hypoventilation and airway spasm is lowered by 10%, that is, the risk threshold for alveolar hypoventilation is adjusted from 60% to 54%, and the risk threshold for airway spasm is adjusted from 50% to 45%; then, combined with the fluctuation amplitude of real-time respiratory electromyography signals, if the fluctuation amplitude is >30μV, the risk warning level is further increased from warning to high warning; the final risk result will indicate the associated risk factors.

8. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The rehabilitation plan generation module implements the following steps: First, the module receives the risk correlation report from the risk assessment module. Combining the patient's postoperative time data and recovery progress data, it determines the core direction of rehabilitation adjustments. If the risk is alveolar hypoventilation and the patient is in the 1-7 day postoperative period, the frequency of airway cleaning training is adjusted from twice a day to four times a day. If the patient is in the 8 days or more postoperative period, the rhythm of breathing training is adjusted from every 20 seconds to every 30 seconds. If the risk is airway spasm, breathing assistance techniques are prioritized, and the rest intervals for breathing training are increased. After the plan is generated, it is first pushed to the management terminal of medical staff, with a 12-hour review window. If the medical staff does not make any modifications, it is automatically pushed to the patient's wearable device, and a check-in reminder for plan execution is generated.

9. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The personalized adaptation requirements of the rehabilitation program generation module also include personalized adaptation steps based on the patient's age and body mass index (BMI). Specifically: if the patient is older than 65 years and has a BMI greater than 28, the intensity of rehabilitation training will be reduced by 20%, while the rest interval for breathing training will be increased from 2 minutes of rest after every 10 minutes of training to 3 minutes of rest after every 8 minutes of training; if the patient is younger than 40 years and has a BMI less than 18, the intensity of rehabilitation training will be increased by 10%, while the number of sets of breathing training will be increased from 3 sets per day to 4 sets per day; in addition, based on the patient's dietary data, if the patient's protein intake is insufficient, guidance on respiratory muscle nutrition supplementation will be added to the rehabilitation program; all adapted and adjusted programs will generate corresponding execution record sheets, recording the reasons for the program adjustments, the content of the adjustments, and the patient's execution status.

10. The intelligent assessment and rehabilitation guidance system for postoperative respiratory function in thoracic surgery according to claim 1, characterized in that, The system also includes a patient feedback module. The specific implementation steps are as follows: the patient provides feedback on their feelings during the rehabilitation training process through voice input on the wearable device or the system's official mini-program; the system converts the patient's feedback data into a quantified feedback score and transmits it synchronously to the rehabilitation plan generation module; if the feedback score is >7, the rehabilitation plan generation module will immediately adjust the intensity and duration of the rehabilitation training.

Images