Rehabilitation equipment integrated management and data interaction system based on mobile terminal

Abstract

The invention belongs to the technical field of complication prevention, and particularly relates to a rehabilitation equipment integrated management and data interaction system based on a mobile terminal, which comprises a multi-source data acquisition module, a data integration and preprocessing module, an intelligent analysis module and a risk early warning module, data islands are cracked through a multi-source heterogeneous data fusion architecture, and an extensible access layer is designed; the system is compatible with medical systems such as HIS, LIS and PACS; a Bluetooth module, a Wi-Fi module and a 5G module are embedded and directly connected with rehabilitation equipment such as a power bicycle and an inspiratory muscle training instrument; cross-institution data collaboration is realized through a federated learning framework, for example, local training sub-models of hospitals and global parameter aggregation of a central server are realized, model generalization is improved on the premise of privacy protection, 'preoperative, intra-operative and post-operative link splitting 'is broken, and a perioperative period full-link data set is constructed.

Description

Technical Field

[0001] The present invention belongs to the technical field of complication prevention, and in particular relates to an integrated management and data interaction system for rehabilitation equipment based on a mobile terminal. Background Art

[0002] Currently, clinical practice generally considers preoperative evaluation, intraoperative monitoring, and postoperative rehabilitation as relatively independent components of the prevention and management of postoperative pulmonary complications (PPCs). During the postoperative period, monitoring of PPCs relies primarily on empirical observations by healthcare professionals and intermittent measurements of physiological indicators. This traditional monitoring approach often struggles to comprehensively and promptly capture early signs of changes in a patient's condition, especially after discharge, due to the lack of effective continuous monitoring and risk warning mechanisms.

[0003] Although technologies such as wearable sensors have been gradually applied to postoperative monitoring in recent years, enabling the initial automated collection of physiological parameters such as heart rate, respiratory rate, and oxygen saturation, these systems typically focus on only a single or a few indicators. For example, the invention patent with publication number CN119479999A achieves automated collection of exercise heart rate and speed data, but its analysis model only revolves around single exercise parameters such as "constant index" and "amplitude assessment." It neither integrates user medical history or environmental data, nor associates with other physiological indicators. There is a lack of effective integration and correlation analysis with multi-dimensional data such as patient medical history, surgical information, and laboratory tests. This makes it difficult for clinicians to obtain a holistic and dynamic assessment of a patient's PPCs risk, resulting in a static and one-sided risk assessment, which is not conducive to early identification of high-risk patients and the implementation of targeted prevention and intervention measures.

[0004] Based on the above background, the present invention proposes an integrated rehabilitation equipment management and data interaction system based on mobile terminals, aiming to break the limitations of traditional monitoring models. By building a platform that can comprehensively integrate various relevant data of patients before, during, and after surgery, and using intelligent analysis methods, early and accurate prediction and continuous monitoring of PPCs risks can be achieved, thereby providing strong support for clinical decision-making and ultimately improving patients' postoperative rehabilitation effects. Summary of the Invention

[0005] In order to solve the above-mentioned problems existing in the prior art, the present invention provides an integrated management and data interaction system for rehabilitation equipment based on mobile terminals, aiming to solve the technical problems of "data silos", "monitoring fragmentation" and "warning lag" in PPCs management in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solutions: a mobile terminal-based integrated management and data interaction system for rehabilitation equipment, which is suitable for comprehensive and accurate assessment and early warning of the risk of postoperative pulmonary complications (PPCs) in patients, and includes a multi-source data acquisition module, a data integration and preprocessing module, an intelligent analysis module, and a risk warning module;

[0007] The multi-source data acquisition module is used to automatically collect data from hospital information systems (HIS), laboratory information systems (LIS), medical imaging systems (PACS), surgical anesthesia systems, surgical scheduling systems, and patient management systems, as well as data from various types of rehabilitation equipment from mobile terminals through standardized interfaces (HL7 / DICOM) and wireless transmission technologies (Bluetooth / Wi-Fi / 4G / 5G);

[0008] The data integration and preprocessing module cleans, structures, and interpolates missing values ​​of multi-source heterogeneous data to generate a unified data set;

[0009] The intelligent analysis module constructs a risk prediction model for postoperative pulmonary complications (PPCs) based on a machine learning algorithm to dynamically assess patient risk;

[0010] The risk warning module sends a multi-channel alert to medical staff via a mobile terminal when the risk score exceeds a threshold;

[0011] In the present invention, multi-source real-time fusion is achieved through a multi-source data acquisition module, a data integration and preprocessing module, an intelligent analysis module, and a risk warning module. Clinical system data such as HIS / LIS / PACS are integrated through standardized interfaces (HL7 / DICOM); Bluetooth / Wi-Fi is used to directly connect mobile terminal rehabilitation equipment (power bicycle / physiological monitor) to break the data barriers between hospital systems and rehabilitation equipment.

[0012] Furthermore, the multi-source data acquisition module at least includes surgical records, test reports and medical images connected to the hospital's HIS, LIS and PACS systems through APIs; the multi-source data acquisition module acquires in real time via Bluetooth or Wi-Fi the power bicycle exercise load data, 6-minute walking distance data, 24-hour physiological monitoring of respiration, heart rate, blood oxygen data and pressure curve data of inspiratory muscle training connected to the mobile terminal;

[0013] In existing technologies, traditional monitoring relies on manual recording, resulting in data fragmentation after discharge, such as the loss of walking test data after discharge. There is also a lack of continuous monitoring in the postoperative period, especially after the patient is discharged.

[0014] In the present invention, the multi-source data acquisition module (API interface, wireless transmission submodule) realizes the full-link coverage of medical treatment and rehabilitation. The API is connected to the surgical anesthesia system to obtain intraoperative parameters (such as tidal volume and oxygenation index); Bluetooth real-time transmission of power bicycle exercise load curve and inspiratory muscle training pressure data covers the entire cycle data of preoperative evaluation, intraoperative monitoring, and postoperative rehabilitation.

[0015] Furthermore, the data integration and preprocessing module uses a time series interpolation algorithm (linear interpolation / spline interpolation) to fill in the missing values ​​of the physiological monitoring data;

[0016] The data integration and preprocessing module interpolates missing values ​​of non-time series data based on random forest or K-nearest neighbor algorithm;

[0017] The data integration and preprocessing module includes time series interpolation and text processing submodules, wherein the text processing submodule applies deepseek-r1 technology to extract structured features from electronic medical record texts;

[0018] In existing technologies, sports systems use simple mean values ​​to fill in missing values, which destroys the temporal correlation of sports data. This makes it difficult for traditional monitoring methods to fully capture disease change signals and leads to data quality issues.

[0019] In the present invention, the data integration and preprocessing module implements dedicated processing of medical time series data, and uses spline interpolation to fill missing values ​​in physiological monitoring data (preserving waveform continuity); deepseek-r1 is used to parse unstructured text such as "pulmonary consolidation" in electronic medical records to solve the data fragmentation problem caused by intermittent measurements.

[0020] Furthermore, the intelligent analysis module integrates a multi-model algorithm, and the multi-model algorithm includes at least a traditional machine learning model and a deep learning model;

[0021] The intelligent analysis module synthesizes the outputs of each model through an integrated learning voting mechanism to generate a dynamic PPCs risk score;

[0022] In existing technologies, CDSS systems are based on static rule bases, such as "alarm if fever > 38.5°C." These systems are unable to learn postoperative physiological decay trends and are unable to obtain a holistic and dynamic assessment of a patient's PPCs risk.

[0023] In the present invention, the intelligent analysis module (LSTM timing engine, integrated voting submodule) realizes medical dynamic risk modeling, LSTM analyzes the variability of postoperative respiratory rate (identifying the periodic decrease in blood oxygen at night); the integrated voting mechanism, the intraoperative parameter weight accounts for 60%, and the rehabilitation data accounts for 40%, thereby realizing dynamic updating of perioperative risks.

[0024] Furthermore, the risk warning module supports custom multi-level risk thresholds;

[0025] The risk warning module sends alerts via at least two of the following methods: mobile application push, SMS, and hospital information system pop-up window;

[0026] The risk warning module automatically optimizes the alarm triggering threshold based on clinical feedback;

[0027] In existing technologies, the sports system generates reports in batches every 2 hours, with a delay of more than 1 hour, making it difficult to fully and timely capture early signals of changes in patients' conditions;

[0028] In the present invention, the risk warning module (threshold dynamic optimization submodule) realizes real-time closed-loop response. When the blood oxygen level drops continuously by more than 5% / 10 minutes, a text message alarm is triggered (delay <10 seconds); the threshold is automatically adjusted according to the false alarm rate (such as the initial threshold is 0.7, which is increased to 0.75 when the false alarm rate is high) to capture early deterioration signals that are easily overlooked by humans.

[0029] Furthermore, the mobile terminal-based integrated rehabilitation equipment management and data interaction system further includes a visualization interaction module, which dynamically displays the patient's physiological indicator trends, rehabilitation training progress, and risk score heat map on the mobile terminal;

[0030] The visual interaction module provides a historical data comparison view to assist medical staff in adjusting rehabilitation plans;

[0031] In existing technologies, CDSS systems only output textual diagnostic suggestions and do not provide data-related views, resulting in information dispersion and a lack of effective continuous monitoring mechanisms.

[0032] In the present invention, the visualization interaction module (heat map engine, trend comparison submodule) realizes the clinical decision-making guidance dashboard, superimposes the risk heat map and the respiratory rate curve (labeled "night frequency increased by 37%"); compares pre-operative / post-operative lung function indicators (such as FEV1 decreased by 20%) to assist manual judgment on the rationality of the model warning.

[0033] Furthermore, the input features of the PPCs risk prediction model include at least the type of surgery, duration of anesthesia, intraoperative blood loss; postoperative respiratory rate variability, frequency of hypoxemia events, inspiratory muscle training tolerance; laboratory inflammatory indicators, and radiographic signs of pulmonary consolidation;

[0034] In existing technologies, general models, such as sports systems, ignore postoperative specific indicators such as "intraoperative blood loss" and lack correlation analysis of multi-dimensional data";

[0035] In the present invention, the intelligent analysis module (medical feature engineering submodule) realizes a postoperative exclusive feature set, constructs cross-features, and calculates the inspiratory muscle training tolerance × CRP inflammatory index; extracts imaging features: the proportion of lung consolidation areas in PACS, thereby improving the model's adaptability to medical scenarios.

[0036] Furthermore, the system supports linkage control of rehabilitation equipment. When the risk score increases, the upper limit of the exercise intensity of the power bicycle is automatically limited; the inspiratory muscle training resistance parameters are dynamically adjusted based on the patient's tolerance;

[0037] In existing technologies, the sports system only recommends "reducing speed" without connecting the device to automatically adjust the speed, which is not conducive to early identification of high-risk patients and taking targeted intervention measures;

[0038] In the present invention, the risk warning module uses risk-driven device intervention:

[0039] if risk_score>0.7:#high risk threshold

[0040] power_bike.set_max_power(50W)#limit the power limit

[0041] insp_trainer.adjust_pressure(curve_slope=0.8)#Adjust the inhalation resistance curve;

[0042] This converts early warning into automatic execution.

[0043] Furthermore, the system adopts a federated learning framework, which at least includes local training of PPCs prediction sub-models in each medical institution and a central server aggregating model parameters to update the global model;

[0044] In existing technologies, CDSS systems avoid cross-institutional data joint modeling due to privacy concerns and lack effective continuous monitoring mechanisms, resulting in cross-hospital data discontinuities after discharge.

[0045] In the present invention, the intelligent analysis module (distributed training submodule) realizes privacy-safe multi-center collaboration. Hospital A locally trains the LSTM submodel → the central server aggregates parameters → updates the global model to break the data compliance barriers.

[0046] Furthermore, the system integrates full-process patient management functions, including at least pushing personalized respiratory rehabilitation training plans through mobile terminals; generating a postoperative follow-up priority list based on risk scores; and automatically generating intervention recommendation reports that comply with clinical guidelines.

[0047] In existing technologies, traditional postoperative management is disconnected, rehabilitation plans are unrelated to real-time risk status, and preoperative, intraoperative, and postoperative stages are considered relatively independent;

[0048] In the present invention, the risk warning module (rehabilitation plan generation submodule) realizes a personalized rehabilitation closed loop, generates a respiratory training plan based on the risk score (such as impedance training three times a day for high-risk patients), automatically pushes it to the mobile terminal and synchronizes it to the inspiratory muscle training instrument parameter library to open up risk management.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] 1. Through a multi-source heterogeneous data fusion architecture, data silos are eliminated and an extensible access layer is designed. The system is compatible with medical systems such as HIS, LIS, and PACS (HL7 and DICOM protocols). Built-in Bluetooth, Wi-Fi, and 5G modules allow for direct connection to rehabilitation equipment such as power bikes and inspiratory muscle trainers. A federated learning framework is used to enable cross-institutional data collaboration, such as training sub-models locally in each hospital and aggregating global parameters on a central server. This improves model generalization while protecting privacy, breaking the "separation between preoperative, intraoperative, and postoperative stages" and constructing a full-link perioperative data set.

[0051] 2. Dynamic time series analysis and real-time early warning are used to resolve monitoring fragmentation, integrating a multi-scale analysis engine. Static data (age, surgery type) are replaced by random forest / GBM classification. Dynamic physiological sequences (respiratory rate, blood oxygen) are replaced by LSTM modeling of attenuation trends. Through a threshold adaptive mechanism, when the risk score exceeds the threshold, SMS and system pop-up alerts are triggered, with a response delay of less than 1 minute. This allows the capture of "early signals that are difficult for humans to perceive" (such as the periodic decrease in blood oxygen at night), providing early warnings 6-8 hours earlier than traditional intermittent monitoring.

[0052] 3. Improve prediction accuracy through multi-model integration and medical scenario optimization; a medical-specific integration strategy that combines traditional models (logistic regression to process structured variables) with deep learning (CNN to analyze PACS images); a dynamic weighted voting mechanism that assigns high weights to intraoperative parameters (such as tidal volume) and downweights post-discharge rehabilitation data; define PPCs key feature sets, such as combining "inspiratory muscle training tolerance" with "inflammatory indicator CRP" to construct cross-features; PPCs prediction AUC is improved and the false alarm rate is reduced.

[0053] 4. Closed-loop management is achieved through the linkage of rehabilitation equipment and personalized intervention, with risk-driven equipment control. When the risk score increases, the resistance limit of the power bicycle is automatically limited (to prevent excessive oxygen consumption), and the inspiratory muscle training parameters (such as the slope of the pressure curve) are dynamically adjusted based on tolerance. The full-process management function automatically generates a respiratory training plan and pushes it to the mobile terminal; the "post-event alarm" is upgraded to "pre-event intervention" to reduce the incidence of complications in medium- and high-risk patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0055] Figure 1 is a flow chart of the present invention;

[0056] Figure 2 This is the overall system architecture diagram of the present invention;

[0057] Figure 3 This is a flowchart of the nighttime warning process in the present invention;

[0058] Figure 4 This is a flowchart of the linkage control of the rehabilitation equipment in the present invention. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] Example 1:

[0061] See also Figures 1-4 This embodiment provides the following technical solutions: a mobile terminal-based integrated management and data interaction system for rehabilitation equipment, which is suitable for comprehensive and accurate assessment and early warning of the risk of postoperative pulmonary complications (PPCs) in patients, and includes a multi-source data acquisition module, a data integration and preprocessing module, an intelligent analysis module, and a risk warning module;

[0062] The multi-source data acquisition module is used to automatically collect data from hospital information systems (HIS), laboratory information systems (LIS), medical imaging systems (PACS), surgical anesthesia systems, surgical scheduling systems, and patient management systems, as well as data from various types of rehabilitation equipment from mobile terminals through standardized interfaces (HL7 / DICOM) and wireless transmission technologies (Bluetooth / Wi-Fi / 4G / 5G). The data integration and preprocessing module cleans, structures, and interpolates missing values ​​for multi-source heterogeneous data to generate a unified data set. The intelligent analysis module constructs a postoperative pulmonary complications (PPCs) risk prediction model based on machine learning algorithms to dynamically assess patient risks. The risk warning module sends multi-channel alerts to medical staff via mobile terminals when the risk score exceeds the threshold.

[0063] Through the multi-source data acquisition module, data integration and preprocessing module, intelligent analysis module, and risk warning module, multi-source real-time fusion is achieved, and clinical system data such as HIS / LIS / PACS are integrated through standardized interfaces (HL7 / DICOM); Bluetooth / Wi-Fi is used to directly connect mobile terminal rehabilitation equipment (power bicycle / physiological monitor) to break the data barriers between hospital systems and rehabilitation equipment.

[0064] The multi-source data acquisition module at least includes surgical records, test reports and medical images connected to the hospital's HIS, LIS and PACS systems through API; the multi-source data acquisition module obtains power bicycle exercise load data, 6-minute walking distance data, 24-hour physiological monitoring respiration, heart rate, blood oxygen data and inspiratory muscle training pressure curve data connected to the mobile terminal in real time through Bluetooth or Wi-Fi; traditional monitoring relies on manual recording, and data is interrupted after discharge, such as the loss of walking test data after discharge. In the postoperative stage, especially after the patient is discharged, there is a lack of continuous monitoring; the multi-source data acquisition module (API interface, wireless transmission sub-module) realizes full medical-rehabilitation link coverage, and the API connects to the surgical anesthesia system to obtain intraoperative parameters (such as tidal volume, oxygenation index); Bluetooth transmits power bicycle exercise load curves and inspiratory muscle training pressure data in real time to cover the full cycle data of preoperative evaluation, intraoperative monitoring and postoperative rehabilitation.

[0065] The data integration and preprocessing module uses a time series interpolation algorithm (linear interpolation / spline interpolation) to fill missing values ​​in physiological monitoring data. The data integration and preprocessing module interpolates missing values ​​in non-time series data based on random forest or K-nearest neighbor algorithms. The data integration and preprocessing module includes time series interpolation and text processing submodules, among which the text processing submodule uses deepseek-r1 technology to extract structured features from electronic medical record texts. The sports system uses a simple mean to fill missing values, which destroys the temporal correlation of sports data, making it difficult for traditional monitoring methods to fully capture disease change signals and causing data quality issues. The data integration and preprocessing module implements dedicated processing of medical time series data and uses spline interpolation to fill missing values ​​in physiological monitoring data (preserving waveform continuity). Deepseek-r1 is used to parse unstructured text such as "pulmonary consolidation" in electronic medical records to solve the problem of data fragmentation caused by intermittent measurements.

[0066] The intelligent analysis module integrates multi-model algorithms, which include at least traditional machine learning models and deep learning models. The intelligent analysis module generates a dynamic PPCs risk score by integrating the outputs of each model through an integrated learning voting mechanism. The CDSS system is based on a static rule base, such as "alarm if fever >38.5°C", and is unable to learn the postoperative physiological attenuation trend, making it difficult to obtain a holistic and dynamic assessment of the patient's PPCs risk. The intelligent analysis module (LSTM timing engine, integrated voting sub-module) implements medical dynamic risk modeling, and LSTM analyzes the variability of postoperative respiratory rate (identifying periodic decreases in blood oxygen at night). The integrated voting mechanism, with intraoperative parameters accounting for 60% of the weight and rehabilitation data accounting for 40%, enables dynamic updates of perioperative risks.

[0067] The risk warning module supports customized multi-level risk thresholds; the risk warning module sends alerts through at least two of the following methods: mobile application push, SMS, and hospital information system pop-up windows; the risk warning module automatically optimizes the alarm trigger threshold based on clinical feedback; the sports system generates reports in batches every 2 hours, with a delay of more than 1 hour, making it difficult to fully and timely capture early signals of changes in the patient's condition; the risk warning module (threshold dynamic optimization submodule) realizes real-time closed-loop response, and triggers SMS alarms (delay <10 seconds) when blood oxygen levels continuously drop by more than 5% / 10 minutes; the threshold is automatically adjusted according to the false alarm rate (such as the initial threshold is 0.7, and it is increased to 0.75 when the false alarm rate is high) to capture early deterioration signals that are easily overlooked by humans.

[0068] The integrated management and data interaction system for rehabilitation equipment based on mobile terminals also includes a visualization interaction module, which dynamically displays the patient's physiological indicator trends, rehabilitation training progress and risk score heat map on the mobile terminal; the visualization interaction module provides a historical data comparison view to assist medical staff in adjusting rehabilitation plans; the CDSS system only outputs text diagnostic suggestions and does not provide a data association view, resulting in information dispersion and a lack of an effective continuous monitoring mechanism; the visualization interaction module (heat map engine, trend comparison sub-module) realizes a clinical decision-oriented dashboard, superimposing the risk heat map and respiratory rate curve (labeled "nighttime frequency increased by 37%"); comparing pre-operative / post-operative lung function indicators (such as a 20% decrease in FEV1) to assist manual judgment on the rationality of model warnings.

[0069] The input features of the PPCs risk prediction model include at least the type of surgery, duration of anesthesia, and intraoperative blood loss; postoperative respiratory rate variability, frequency of hypoxemia, and tolerance to inspiratory muscle training; laboratory inflammatory indicators, and radiographic signs of pulmonary consolidation. General models, such as the sports system, ignore postoperative-specific indicators such as intraoperative blood loss and lack correlation analysis of multidimensional data. The intelligent analysis module (medical feature engineering submodule) implements a postoperative-specific feature set and constructs cross-features, such as tolerance to inspiratory muscle training × CRP inflammatory index. Imaging features are extracted, including the proportion of pulmonary consolidation areas in PACS, thereby improving the model's adaptability to medical scenarios.

[0070] The system supports the linkage control of rehabilitation equipment. When the risk score increases, the upper limit of the exercise intensity of the power bicycle is automatically limited; the inspiratory muscle training resistance parameters are dynamically adjusted based on the patient's tolerance; the sports system only recommends "reducing speed" and does not connect the equipment for automatic speed adjustment, which is not conducive to early identification of high-risk patients and targeted intervention measures; the risk warning module uses risk-driven equipment intervention to convert the warning into automatic execution.

[0071] The system adopts a federated learning framework, which at least includes local training of PPCs prediction sub-models in each medical institution and central server aggregation of model parameters to update the global model; the CDSS system avoids cross-institutional data joint modeling due to privacy concerns and lacks an effective continuity monitoring mechanism, resulting in cross-hospital data disconnection after discharge; the intelligent analysis module (distributed training sub-module) realizes privacy-safe multi-center collaboration, where Hospital A locally trains the LSTM sub-model → central server aggregates parameters → updates the global model to break through data compliance barriers.

[0072] The system integrates comprehensive patient management capabilities, including at least personalized respiratory rehabilitation training plans pushed to mobile devices; a postoperative follow-up priority list generated based on risk scores; and automatic generation of intervention recommendation reports that align with clinical guidelines. Traditional postoperative management is disconnected, with rehabilitation plans unrelated to real-time risk status, and preoperative, intraoperative, and postoperative phases treated as relatively independent. The risk warning module (rehabilitation plan generation submodule) implements a personalized rehabilitation closed-loop, generating respiratory training plans based on risk scores (e.g., three daily impedance training sessions for high-risk patients). These plans are automatically pushed to mobile devices and synchronized with the inspiratory muscle training instrument parameter library, thus facilitating risk management.

[0073] Example 2:

[0074] See also Figures 1-4In this embodiment, staff, using the disclosed system, performed a thoracoscopic right lobectomy in the thoracic surgery ward of a tertiary hospital. A 65-year-old patient, surnamed Zhang, had just completed a thoracoscopic right lobectomy. The patient's bedside multi-parameter monitor, smart bracelet, and mobile tablet were all connected to the hospital's integrated rehabilitation equipment management and data exchange system. One day before surgery, the system automatically retrieved the patient's COPD medical history and pulmonary function data (FEV1 = 65%) from the HIS, and obtained a chest CT image (a 2cm right lung nodule) from the PACS. The surgical scheduling system then marked the procedure as a thoracoscopic lobectomy.

[0075] During the operation, the anesthesia system uploaded real-time data showing the patient's oxygenation index dropping to 92% and blood loss reaching 150ml. On the first postoperative day, the smart bracelet recorded an average respiratory rate of 22 breaths / minute every five minutes, and blood oxygen remained at 96%. On the third postoperative day, while the patient was training on a Bluetooth-connected power cycle, the system recorded a peak power of 60W for 10 minutes. On the fifth postoperative day, the inspiratory muscle training device uploaded a pressure curve showing a maximum inspiratory pressure of 45cmH2O.

[0076] A key event occurred at 2:15 AM on the second day after surgery: the smart bracelet detected a spike in the patient's respiratory rate from 20 breaths / minute to 28 breaths / minute, while his blood oxygen level fluctuated from 96% to 90%. This state persisted for 10 minutes before partially recovering. The system immediately triggered an analysis process: the LSTM model identified the "periodic drop in blood oxygen level during the night," and the random forest model associated it with intraoperative hypoxic events, assigning a 30% weight. Integrated voting generated a risk score of 0.68 (exceeding the preset threshold of 0.65).

[0077] Within three seconds, a red alert popped up on the bedside tablet: "High-risk alert! Suspected early-stage atelectasis. Recommendations: ① Urgent lung auscultation ② Enhanced airway clearance training." A text message was simultaneously sent to the on-call doctor's mobile phone. The doctor rushed to the ward, auscultated, and confirmed diminished breath sounds at the right lung base. Bronchoscopy and suctioning were immediately arranged. Simultaneously, the system automatically activated device linkage: the power bike's maximum power was limited to 40W (from 60W), and the inspiratory muscle training device's resistance was reduced by 20%. The next day, the mobile terminal pushed an adjustment plan: "Add two high-frequency chest wall oscillation treatments daily, and prioritize chest X-ray review."

[0078] A CT scan on the seventh day after surgery revealed no complications, and the patient was discharged after eight days of hospitalization (compared to the typical average of 12 days). The system integrates surgical data, real-time physiological indicators, and rehabilitation equipment parameters to achieve closed-loop management from risk warning to intervention execution.

[0079] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A mobile terminal-based integrated rehabilitation equipment management and data interaction system. This mobile terminal-based rehabilitation equipment integrated management and data interaction system is suitable for comprehensive and accurate assessment and early warning of the risk of postoperative pulmonary complications (PPCs) in patients. It features: It includes multi-source data acquisition module, data integration and pre-processing module, intelligent analysis module and risk warning module; The multi-source data acquisition module is used to automatically collect data from hospital information systems (HIS), laboratory information systems (LIS), medical imaging systems (PACS), surgical anesthesia systems, surgical scheduling systems, and patient management systems, as well as data from various types of rehabilitation equipment from mobile terminals through standardized interfaces (HL7 / DICOM) and wireless transmission technologies (Bluetooth / Wi-Fi / 4G / 5G); The data integration and preprocessing module cleans, structures, and interpolates missing values ​​of multi-source heterogeneous data to generate a unified data set; The intelligent analysis module constructs a risk prediction model for postoperative pulmonary complications (PPCs) based on a machine learning algorithm to dynamically assess patient risk; When the risk score exceeds a threshold, the risk warning module sends a multi-channel alert to medical staff via a mobile terminal.

2. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: The multi-source data acquisition module at least includes surgical records, test reports and medical images connected to the hospital HIS, LIS and PACS systems through API; the multi-source data acquisition module obtains in real time through Bluetooth or Wi-Fi the power bicycle exercise load data, 6-minute walking distance data, 24-hour physiological monitoring of breathing, heart rate, blood oxygen data and pressure curve data of inspiratory muscle training connected to the mobile terminal.

3. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: The data integration and preprocessing module uses a time series interpolation algorithm (linear interpolation / spline interpolation) to fill in missing values ​​of physiological monitoring data; The data integration and preprocessing module interpolates missing values ​​of non-time series data based on random forest or K-nearest neighbor algorithm; The data integration and preprocessing module includes time series interpolation and text processing submodules, wherein the text processing submodule uses deepseek-r1 technology to extract structured features from electronic medical record texts.

4. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: The intelligent analysis module integrates a multi-model algorithm, and the multi-model algorithm includes at least a traditional machine learning model and a deep learning model; The intelligent analysis module integrates the outputs of each model through an integrated learning voting mechanism to generate a dynamic PPCs risk score.

5. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: The risk warning module supports custom multi-level risk thresholds; The risk warning module sends alerts via at least two of the following methods: mobile application push, SMS, and hospital information system pop-up window; The risk warning module automatically optimizes the alarm triggering threshold based on clinical feedback.

6. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: It also includes a visualization interaction module, which dynamically displays the patient's physiological indicator trends, rehabilitation training progress and risk score heat map on the mobile terminal; The visual interaction module provides a historical data comparison view to assist medical staff in adjusting rehabilitation plans.

7. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: The input features of the PPCs risk prediction model include at least the type of surgery, duration of anesthesia, and intraoperative blood loss; postoperative respiratory rate variability, frequency of hypoxemia events, and tolerance to inspiratory muscle training; laboratory inflammatory indicators, and radiographic signs of pulmonary consolidation.

8. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: The system supports linkage control of rehabilitation equipment: When the risk score increases, the upper limit of the exercise intensity of the power bicycle will be automatically limited; Dynamically adjust inspiratory muscle training resistance parameters based on patient tolerance.

9. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: The system adopts a federated learning framework, which at least includes local training of PPCs prediction sub-models in each medical institution and a central server aggregating model parameters to update the global model.

10. The mobile terminal-based integrated rehabilitation equipment management and data interaction system according to claim 1, characterized in that: The system integrates the patient management function throughout the entire process, including at least: Push personalized respiratory rehabilitation training plans through mobile terminals; Generate a postoperative follow-up priority list based on the risk score; Automatically generate reports with recommended interventions that comply with clinical guidelines.

Images