Portable respiratory function rehabilitation training device and method thereof

Abstract

The invention discloses a portable respiratory function rehabilitation training device and a method thereof, and belongs to the technical field of medical rehabilitation. The device comprises a data acquisition module which integrates a plurality of micro sensors to acquire respiratory airflow, heart rate, oxyhemoglobin saturation and environmental parameters of a user in real time; the data preprocessing and fusion module is used for cleaning, standardizing and performing time synchronization on the collected original data and fusing the collected original data into a uniform data format; the intelligent analysis module is responsible for performing real-time analysis on the fused data, identifying a breathing mode, evaluating a current health state and predicting a rehabilitation trend; the training scheme generation module is used for dynamically generating and adjusting a personalized breathing training scheme; the real-time feedback and interaction module is used for providing real-time training guidance and progress feedback for the user; the effect evaluation and optimization module is used for evaluating the rehabilitation effect and continuously optimizing a training strategy by using a reinforcement learning algorithm; according to the portable hardware integration module, all modules are integrated into small-sized and low-power-consumption portable equipment.

Description

Technical Field

[0001] This application relates to the field of medical rehabilitation technology, and more specifically, to a portable breathing function rehabilitation training device and its method. Background Art

[0002] In the field of respiratory medicine, respiratory monitoring is a key link in the management and treatment of various respiratory diseases. With the progress of medical technology, although respiratory monitoring devices are constantly updated, many challenges still remain. Traditional respiratory monitoring methods usually rely on large-scale devices in hospitals, which are usually bulky and fixedly installed, limiting their flexibility and convenience in clinical settings. Although such devices provide comprehensive physiological data monitoring, they are insufficient in supporting patients' daily rehabilitation and long-term health management.

[0003] Although current respiratory monitoring devices have made progress in data acquisition and analysis, many devices still have deficiencies such as single functions, limited data processing capabilities, lack of personalized and intelligent adjustments. These devices often cannot provide comprehensive health assessments and dynamic adjustment functions in real time, resulting in patients being unable to obtain accurate rehabilitation guidance and immediate feedback in practical applications.

[0004] Although traditional portable respiratory monitoring devices have improved in mobility, they still face the problem of integrating multiple monitoring functions. Most portable devices can only perform basic respiratory parameter monitoring and lack the ability to comprehensively analyze complex physiological data and generate personalized training programs. In addition, the data processing and feedback mechanisms in existing devices usually cannot dynamically adapt to patients' health changes, resulting in inaccurate evaluation and adjustment of training effects.

[0005] In summary, how to implement a portable, multi-functional, intelligent analysis and personalized adjustment capable respiratory monitoring device to provide comprehensive support in the clinical applications of respiratory medicine has become a technical problem that urgently needs to be solved. Summary of the Invention

[0006] In order to overcome a series of defects existing in the prior art, the purpose of this application is to provide a portable breathing function rehabilitation training device for the above problems, including the following modules:

[0007] A data acquisition module that integrates multiple micro sensors to collect the user's respiratory airflow, heart rate, blood oxygen saturation, and environmental parameters in real time;

[0008] A data preprocessing and fusion module that cleans, standardizes, and synchronizes the time of the collected raw data and fuses it into a unified data format;

[0009] The intelligent analysis module is responsible for real-time analysis of the fused data, identifying breathing patterns, evaluating the current health status, and predicting the rehabilitation trend;

[0010] The training plan generation module dynamically generates and adjusts personalized breathing training plans based on the intelligent analysis results and preset training goals to ensure that the training content always matches the user's real-time status and needs;

[0011] The real-time feedback and interaction module provides instant training guidance and progress feedback to the user through visual, auditory, and tactile means;

[0012] The effect evaluation and optimization module continuously records and analyzes the user's long-term training data, evaluates the rehabilitation effect, and uses reinforcement learning algorithms to continuously optimize the training strategy to achieve personalized long-term health management;

[0013] The portable hardware integration module integrates all modules into a small-sized, low-power portable device to enable training anytime and anywhere.

[0014] Furthermore, the data acquisition module includes the following components:

[0015] The airflow sensor unit measures the user's breathing airflow rate and breathing frequency in real time to provide basic data for analyzing breathing patterns and intensity;

[0016] The photoplethysmography sensor unit measures the user's heart rate and blood oxygen saturation simultaneously through optoelectronic technology;

[0017] The environmental sensor unit monitors the temperature, humidity, and air quality parameters of the training environment in real time to ensure the comfort and safety of the training environment;

[0018] The attitude sensor unit uses an accelerometer and a gyroscope to collect the user's attitude information in real time to ensure that breathing training is carried out in the correct state;

[0019] The body temperature sensor unit is used to detect the user's skin surface temperature and evaluate the user's body temperature changes during training in real time.

[0020] Furthermore, the data preprocessing and fusion module includes the following components:

[0021] The data reception and caching unit receives various raw data from the data acquisition module and performs temporary storage;

[0022] The data cleaning unit performs outlier detection, denoising, and interpolation on the raw data to improve the data quality and reliability;

[0023] The data standardization unit converts data from different sources and scales to a unified standard range for subsequent analysis and comparison;

[0024] The time synchronization unit aligns the data from different sensors to ensure the consistency of the data in the time dimension;

[0025] The data fusion unit integrates the multi-source data after cleaning, standardization and synchronization to generate a unified data structure;

[0026] The output formatting unit converts the fused data into a predefined standard format for the subsequent modules to process and analyze.

[0027] Furthermore, the intelligent analysis module includes the following components:

[0028] The feature extraction unit extracts key features from the fused data to provide high-quality input for subsequent analysis;

[0029] The deep learning model unit uses a pre-trained neural network model to deeply analyze the extracted features and identify complex data patterns;

[0030] The time series analysis unit uses time series analysis techniques to capture the time dependence and change trends of the data;

[0031] The breathing pattern recognition unit combines the results of deep learning and time series analysis to identify and classify different breathing patterns;

[0032] The health status assessment unit evaluates the user's current health status based on the identified breathing patterns and other physiological indicators;

[0033] The rehabilitation trend prediction unit integrates the user's historical health data and the current assessment results to predict the user's rehabilitation trend and potential risks.

[0034] Furthermore, the training plan generation module includes the following components:

[0035] The user profile management unit stores and updates the user's long-term health data, training history and personal preferences to provide background information for plan generation;

[0036] The training goal setting unit sets and manages short-term and long-term training goals according to medical advice and user wishes;

[0037] The real-time status input unit receives and processes the real-time user status data from the intelligent analysis module to ensure that the plan matches the current situation;

[0038] The training plan generator comprehensively considers the user profile, training goals and real-time status to generate a preliminary personalized training plan;

[0039] The plan adaptability adjustment unit dynamically adjusts and optimizes the training plan according to the user's real-time feedback and training progress to ensure its continuous effectiveness;

[0040] The training instruction output unit converts the generated training plan into specific and executable training instructions and transmits them to the real-time feedback and interaction module.

[0041] Furthermore, the real-time feedback and interaction module includes the following components:

[0042] The training instruction receiving unit receives the specific instructions from the training plan generation module and provides the basic content for feedback generation;

[0043] The multimodal feedback generator generates feedback content suitable for different sensory channels according to the training instructions and the user's real-time status;

[0044] The visual feedback unit displays visual information including training guidance, progress bars, and performance statistics through a graphical interface;

[0045] The auditory feedback unit uses voice prompts and sound effects to provide real-time training guidance and encouraging feedback;

[0046] The tactile feedback unit provides training rhythm guidance and completion prompts through tactile signals such as vibration;

[0047] The user input processing unit receives and parses the user's operations and voice instructions to achieve two-way interaction.

[0048] Furthermore, the effect evaluation and optimization module includes the following components:

[0049] The long-term data recording unit continuously collects and stores the user's training data, physiological indicators, and health status, providing comprehensive historical data support for subsequent analysis;

[0050] The effect evaluation engine comprehensively analyzes the user's rehabilitation progress and training effect based on the recorded long-term data and generates a quantitative evaluation report;

[0051] The health trend analysis unit uses time series analysis techniques to deeply analyze the long-term changes in the user's health status and identify key health trends and patterns;

[0052] The reinforcement learning optimizer models the training process as a reinforcement learning problem and continuously optimizes the training strategy based on the evaluation results and health trends;

[0053] The policy update unit receives the optimized policy and converts it into specific training plan adjustment suggestions to achieve dynamic update of the training strategy;

[0054] The personalized parameter adjustment unit fine-tunes the optimized training strategy according to the user's individual characteristics and feedback to ensure its adaptability and personalization.

[0055] The object of the present application is also to provide a portable breathing function rehabilitation training method, which is implemented based on the above-mentioned portable breathing function rehabilitation training device, and includes the following steps:

[0056] Real-time collect the user's physiological parameters, posture information and environmental parameters;

[0057] Clean, standardize and synchronize the collected data in time, and integrate them into a unified format;

[0058] Extract the key features of the user's breathing function from the integrated data, identify the breathing pattern, and conduct real-time health status assessment;

[0059] Generate a preliminary personalized training plan according to the user's personal profile, training objectives and real-time health status;

[0060] Provide instant training guidance for the user through multi-modal feedback and feedback the current progress;

[0061] Continuously record the user's long-term training data and generate a quantitative training effect evaluation report;

[0062] According to the evaluation results, update and fine-tune the training strategy to achieve adaptive optimization and personalized iteration of the training plan.

[0063] Compared with the prior art, the present application has the following beneficial effects:

[0064] The present application integrates a variety of micro sensors, data preprocessing, intelligent analysis, personalized training plan generation and real-time feedback modules, can collect the user's physiological parameters in real time and analyze the breathing pattern through a deep learning model, dynamically generate a personalized rehabilitation training plan, and guide the user through multi-modal feedback, continuously optimize the training strategy, and realize intelligent and personalized breathing rehabilitation training and long-term health management. Description of the Drawings

[0065] Figure 1 It is a schematic structural diagram of a portable breathing function rehabilitation training device disclosed in an embodiment of the present application;

[0066] Figure 2 It is a flowchart of a portable breathing function rehabilitation training method disclosed in an embodiment of the present application. Detailed Embodiments

[0067] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be described in more detail below with reference to the drawings in the embodiments of the present invention. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some, but not all, of the embodiments of the present invention.

[0068] Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0069] The embodiments described below with reference to the accompanying drawings and directional terms are exemplary and are intended to explain the present invention, and should not be construed as a limitation to the present invention.

[0070] As Figure 1 shown, a portable breathing function rehabilitation training device includes the following modules:

[0071] A data acquisition module that integrates a variety of micro sensors to collect the user's breathing airflow, heart rate, blood oxygen saturation and environmental parameters in real time;

[0072] A data preprocessing and fusion module that cleans, standardizes and synchronizes the time of the collected raw data, and fuses them into a unified data format;

[0073] An intelligent analysis module responsible for analyzing the fused data in real time, identifying breathing patterns, evaluating the current health status, and predicting the rehabilitation trend;

[0074] A training plan generation module that dynamically generates and adjusts personalized breathing training plans based on the intelligent analysis results and preset training goals to ensure that the training content always matches the user's real-time status and needs;

[0075] A real-time feedback and interaction module that provides instant training guidance and progress feedback to the user through visual, auditory and tactile means;

[0076] An effect evaluation and optimization module that continuously records and analyzes the user's long-term training data, evaluates the rehabilitation effect, and uses reinforcement learning algorithms to continuously optimize the training strategy to achieve personalized long-term health management;

[0077] A portable hardware integration module that integrates all modules into a small-sized and low-power portable device to enable training anytime and anywhere.

[0078] In this embodiment, the data acquisition module is one of the core components of the portable respiratory function rehabilitation training device. It integrates a variety of micro sensors to collect the user's respiratory airflow, heart rate, blood oxygen saturation, and environmental parameters in real time. The respiratory airflow sensor uses thermocouple or pressure sensor technology to accurately measure the user's respiratory rate and depth. The heart rate sensor is generally based on the principle of photoplethysmography (PPG), and calculates the heart rate by detecting the minute changes in blood flow. The blood oxygen saturation is monitored by spectroscopic methods to ensure that the user's oxygen supply is within the normal range. The collection of environmental parameters, such as temperature, humidity, and air pressure, helps to provide a comprehensive background for health assessment. These sensors need to have high sensitivity and stability to ensure the accuracy and reliability of the data, especially in cases where there are large changes in movement or breathing patterns.

[0079] In this embodiment, the data preprocessing and fusion module is responsible for cleaning, standardizing, and time-synchronizing the raw data obtained from each sensor to ensure the accuracy and consistency of the data. The data cleaning process includes removing noise, filling in missing values, and correcting outliers to improve the quality of the data. The standardization process converts data from different sources into a unified unit and scale to make them comparable. Time synchronization is a key step in processing dynamic data, ensuring the consistency of all data streams in time, so that subsequent analysis can be based on an accurate time series. Data fusion integrates the cleaned and standardized data into a unified format for further intelligent analysis. Efficient data preprocessing and fusion can significantly improve the system's response speed and analysis accuracy.

[0080] In this embodiment, the intelligent analysis module is the "brain" of the portable respiratory function rehabilitation training device. It is responsible for real-time analysis of the fused data, identifying the user's breathing pattern, evaluating the current health status, and predicting the rehabilitation trend. This module usually uses machine learning and data mining techniques to train on a large amount of historical data to identify normal and abnormal breathing patterns. Through pattern recognition, it can detect whether the user's breathing is regular and whether there are breathing difficulties or other health problems. The health status assessment comprehensively evaluates the user's overall health level by analyzing parameters such as heart rate and blood oxygen. Predicting the rehabilitation trend is based on the existing data and models to estimate future health changes, providing a basis for formulating personalized training programs. The accuracy and real-time nature of intelligent analysis directly affect the effectiveness of the entire system.

[0081] In this embodiment, the training plan generation module dynamically generates and adjusts a personalized breathing training plan based on the intelligent analysis results and preset training goals. This module uses algorithms to transform the user's current health status and rehabilitation goals into specific training tasks and plans. According to the analysis results, suitable breathing exercises, training intensities, and durations can be recommended. In addition, this module can also adjust the training plan in real time in response to changes in the user's state during training, such as changes in breathing rate and fluctuations in heart rate. The personalized training plan can improve the effectiveness of training and reduce the risks caused by mismatched training for the user. The dynamic adjustment function ensures that the training content always matches the user's real-time needs, enhancing the rehabilitation effect.

[0082] In this embodiment, the real-time feedback and interaction module includes a user interface that conveys training information through graphical interfaces, sound prompts, and vibration feedback. Visual feedback includes charts, animations, and real-time data displays to help users understand their training progress and effects. Auditory feedback guides users to perform correct breathing exercises through voice prompts or sound signals. Tactile feedback can remind users to adjust their postures or breathing rhythms through vibration devices. Real-time feedback not only improves the user's sense of participation in training but also helps users correct mistakes in a timely manner and optimize the training effect.

[0083] In this embodiment, through long-term data collection and analysis, it is possible to evaluate the user's rehabilitation progress and identify the successful factors and deficiencies in training. The reinforcement learning algorithm optimizes the training plan through continuous feedback and adjustment to improve the personalization and effectiveness of training. The core of this module lies in continuously improving the training strategy through data-driven optimization to make it more in line with the user's needs and support long-term health management.

[0084] In this embodiment, the design of the portable hardware integration module needs to take into account the volume, weight, and battery life of the device to ensure the portability and comfort of the device during actual use. The integrated hardware usually includes a processor, sensors, a battery, and a display screen, and all components need to operate efficiently within a limited space. Low-power design is an important factor to ensure the long-term use of the device, and at the same time, the heat dissipation and stability of the device need to be considered. Good hardware integration not only improves the convenience of using the device but also enhances the overall user experience.

[0085] In summary, the portable respiratory function rehabilitation training device disclosed in this embodiment realizes efficient and personalized rehabilitation training through the collaboration of its various modules. The data acquisition module provides accurate physiological and environmental data, and the data preprocessing and fusion module ensures the consistency and accuracy of the data. The intelligent analysis module performs real-time analysis and health assessment of the data through advanced algorithms, and the training program generation module dynamically adjusts the training content according to the analysis results. The real-time feedback and interaction module enhances the user's training experience and effect, and the effect evaluation and optimization module improves the training strategy through data-driven optimization. Finally, the portable hardware integration module integrates these functions into an easy-to-carry device, allowing users to perform rehabilitation training anytime, anywhere. Overall, the device provides a comprehensive and personalized solution for respiratory function rehabilitation.

[0086] Furthermore, the data acquisition module includes the following components:

[0087] Airflow sensor unit, which measures the user's breathing airflow rate and breathing frequency in real time, providing basic data for analyzing breathing patterns and intensity;

[0088] A photoplethysmography sensor unit that simultaneously measures the user’s heart rate and blood oxygen saturation using photoelectric technology;

[0089] Environmental sensor unit, which monitors the temperature, humidity and air quality parameters of the training environment in real time to ensure the comfort and safety of the training environment;

[0090] The posture sensor unit uses an accelerometer and a gyroscope to collect the user's posture information in real time to ensure that breathing training is performed in the correct state;

[0091] The body temperature sensor unit is used to detect the user's skin surface temperature and evaluate the user's body temperature changes during training in real time.

[0092] In this embodiment, the airflow sensor unit is one of the key components of the data acquisition module, and its main function is to measure the user's breathing airflow rate and breathing frequency in real time. These data are crucial for analyzing the user's breathing pattern and intensity. The airflow sensor generally uses thermocouple or piezoelectric sensor technology to measure the breathing rate by sensing changes in air flow. These sensors can record the airflow volume of each breath with high precision, thereby providing detailed breathing intensity data. The breathing frequency is determined by counting the number of breaths per unit time. High-precision airflow sensors can accurately capture tiny breathing changes and help identify whether there is dyspnea or abnormal breathing patterns. These data are not only used for real-time monitoring, but also provide important basic data for subsequent intelligent analysis, which is convenient for formulating personalized training programs suitable for users.

[0093] In this embodiment, the photoplethysmography sensor unit (PPG sensor unit) uses optoelectronic technology to simultaneously measure the user's heart rate and blood oxygen saturation. The PPG sensor can calculate the heart rate and blood oxygen level by emitting light and detecting the changes in the absorption and reflection of light in the blood. The measurement of the heart rate is based on the light intensity fluctuations caused by blood flow, which are synchronized with the heartbeat, thus enabling real-time monitoring of the heart rate. The measurement of blood oxygen saturation is determined by analyzing the absorption difference between red light and infrared light to determine the oxygen content in the blood. This sensor has the characteristics of non-invasiveness and strong real-time performance, enabling the user to continuously monitor the heart rate and blood oxygen level during training to ensure training within a healthy range. These data are of great significance for evaluating the user's overall health status and training effect.

[0094] In this embodiment, the environmental sensor unit is responsible for real-time monitoring of the temperature, humidity, and air quality parameters of the training environment to ensure the comfort and safety of the training environment. The temperature and humidity sensor provides basic climate information of the environment by measuring the air temperature and humidity, which is crucial for evaluating the user's training comfort and breathing conditions. For example, high temperature or high humidity may affect the user's breathing efficiency and cause discomfort. The air quality sensor monitors the concentration of pollutants in the environment, such as carbon dioxide, volatile organic compounds, etc., to ensure that the air quality of the training environment meets the health standards. These environmental parameter data help to adjust the training intensity and content, or remind the user to train in an unsuitable environment, enhancing the safety and effectiveness of training.

[0095] In this embodiment, the attitude sensor unit uses an accelerometer and a gyroscope to collect the user's attitude information in real time to ensure that the user performs breathing training in the correct state. The accelerometer can measure the acceleration of the user in different directions to determine the body tilt and motion state of the user. The gyroscope measures the angular velocity of the user to help monitor the rotation and attitude changes of the body. By combining the data of these two sensors, it is possible to accurately judge whether the user's attitude meets the training requirements and provide attitude adjustment suggestions. This function can prevent the risk of poor training effect or injury caused by incorrect attitude of the user, improving the safety and effectiveness of training.

[0096] In this embodiment, the body temperature sensor unit is used to detect the skin surface temperature of the user and evaluate the body temperature change of the user during training in real time. The body temperature change can reflect the user's physical state, such as whether there is overheating, excessive sweating, or other abnormal conditions. Common body temperature sensors include thermocouples or infrared sensors, which can accurately record the body surface temperature to help monitor the possible body temperature abnormalities of the user during training. These data are of great significance for evaluating the training intensity, adjusting the training plan, and preventing health problems such as overheating. Real-time body temperature monitoring can help the system timely adjust the training plan to ensure that the user trains within a safe temperature range.

[0097] In summary, the various components of the data acquisition module work together to ensure that the portable breathing function rehabilitation training device can accurately and comprehensively monitor the user's physiological state and training environment. The airflow sensor unit provides basic data on breathing airflow and frequency, while the photoplethysmography sensor unit monitors heart rate and blood oxygen saturation in real time. The environmental sensor unit ensures the comfort and safety of the training environment, and the attitude sensor unit ensures that the user performs training in the correct posture. The body temperature sensor unit monitors changes in body temperature to provide further health assessments. Overall, through precise data acquisition and real-time monitoring, these components provide a solid foundation for intelligent analysis and the formulation of training plans, thereby improving the personalization, effectiveness, and safety of rehabilitation training.

[0098] Furthermore, the acceleration sensor and gyroscope are used to collect the user's attitude information in real time to ensure that breathing training is carried out in the correct state, including the following steps:

[0099] The acceleration data of the user's body is collected in real time by the acceleration sensor for perceiving the position changes of the user in each axial direction; at the same time, the angular velocity data of the user's body is captured by the gyroscope to obtain the changes in its attitude angle. The acceleration sensor measures the acceleration data of the user's body in three axial directions, namely the x-axis (front-back direction), y-axis (left-right direction), and z-axis (up-down direction). These acceleration data can help perceive the position changes of the user in these directions, thereby inferring the user's posture and motion state. The gyroscope measures the angular velocity of the user's body and captures the rotation of the body in each direction. These angular velocity data are used to calculate the changes in the attitude angle, such as the pitch angle, roll angle, and yaw angle of the body. By combining these data, the real-time attitude information of the user can be obtained.

[0100] The data collected by the acceleration sensor and gyroscope are fused and processed to calculate the actual attitude angle and direction of the user. Since there are certain errors and noises in the measurements of the acceleration sensor and gyroscope respectively, data fusion is a key step to improve the accuracy of attitude estimation. Common methods include the Kalman filter or complementary filter, which can integrate the static information of the acceleration sensor and the dynamic information of the gyroscope to provide a more stable and accurate attitude estimation. The Kalman filter uses a dynamic model and noise characteristics to fuse sensor data, while the complementary filter improves the reliability of attitude estimation by weighted averaging of low-frequency and high-frequency signals. Through data fusion processing, the actual attitude angle of the user can be obtained, including the tilt angle and direction of the body.

[0101] By comparing the user's current posture with a preset ideal training posture in real time, it is ensured that the user performs breathing training in the correct posture, and a prompt signal is generated when a deviation from the standard posture is detected. After collecting and processing the user's posture data in real time, the current posture is compared with the preset ideal training posture. These ideal postures are usually standard postures set based on the best training effect or safety considerations. By calculating the deviation between the current posture and the ideal posture, it is determined whether the user is in the correct training state. When it is detected that the user's posture deviates from the standard, a prompt signal will be generated. These prompt signals can be visual prompts (such as diagrams or animations on the display screen), auditory prompts (such as voice prompts or sound alarms), or tactile prompts (such as vibration feedback). These prompts help the user adjust the posture in time to ensure the correctness and effectiveness of the training.

[0102] Based on a preset posture threshold judgment criterion, potential abnormal postures during the training process are identified, and the user is prompted to correct them through multimodal feedback to ensure the effectiveness and safety of the training. For example, if the user's posture exceeds the set allowable range, it will be determined as an abnormal posture. To further improve the training effect and safety, a multimodal feedback mechanism is used to prompt the user to correct the posture. Multimodal feedback includes visual feedback, auditory feedback, and tactile feedback, providing correction suggestions to the user in multiple ways. Visual feedback may be manifested as posture adjustment prompts on the screen, auditory feedback may include voice instructions, and tactile feedback may include vibration prompts. This multimodal feedback mechanism can help the user quickly perceive and correct abnormal postures, thereby avoiding potential risks during the training process and improving the overall training effect.

[0103] In summary, the accurate collection and processing of posture information are crucial for the effectiveness and safety of breathing training. By using an accelerometer and a gyroscope to collect the user's acceleration and angular velocity data in real time and performing data fusion processing, the user's true posture information can be obtained. Comparing this information with the ideal training posture ensures that the user performs training in the correct posture, and the prompt signal helps the user adjust the posture in time. Abnormal detection and correction are performed based on the preset posture threshold to further ensure the effectiveness and safety of the training. Overall, these steps not only improve the personalization and accuracy of breathing training but also enhance the user's training experience and effect.

[0104] Furthermore, the data collected by the accelerometer and the gyroscope are fused and processed to calculate the user's actual posture angle and direction, including the following steps:

[0105] In a static posture, use the accelerometer to measure the initial angle θ 0. By measuring the acceleration components exerted on the sensor by gravity, the accelerometer can provide angle information related to the direction of gravity. Specifically, the accelerometer can measure the tilt angle of the sensor relative to the direction of gravity, which is the basic angle in the static posture. This angle θ 0 will be used as the initial value for subsequent attitude estimation and fused with dynamic attitude data;

[0106] Within each time step Δt, the angular velocity ω measured by the gyroscope r updates the angle estimate, and the calculation formula is: θ g,r = θ r-1 + ω r ·Δt, to predict the current attitude angle θ g,r , using the angle θ r-1 at the previous moment and the angular velocity ω r ;

[0107] Fuse the gyroscope angle estimate θ g,r and the accelerometer measurement angle θ a,r to calculate the final attitude angle θ r , that is, the actual posture and direction of the user. The calculation formula is: θ r = α·θ g,r + (1 - α)·θ a,r , where α is the filtering factor, ranging from 0 to 1. This filtering factor α determines the weights of the gyroscope data and the accelerometer data in the final angle calculation. A larger α value indicates more reliance on the gyroscope data, and a smaller α value indicates more reliance on the accelerometer data. Through this weighted fusion, the advantages of the two sensors can be combined to obtain a more accurate and stable attitude angle. In addition, the calculated final attitude angle θ r represents the actual attitude angle and direction of the user. This result combines the stability of the static accelerometer measurement and the timeliness of the dynamic gyroscope data. The final attitude angle can be used for further attitude comparison, correction, and optimization to ensure the correct posture of the user during training.

[0108] In summary, the fusion processing steps of the attitude angle achieve an accurate estimation of the user's actual attitude by combining the data of the accelerometer and the gyroscope. Measuring the initial angle in the static attitude provides a basis for subsequent dynamic updates. Predicting the angle through the angular velocity data of the gyroscope, combining the stable measurement results of the accelerometer, and using the filtering factor for data fusion to obtain the final attitude angle. Such a method not only improves the accuracy and stability of attitude estimation but also provides reliable data support for real-time attitude monitoring and training feedback.

[0109] Furthermore, the data preprocessing and fusion module includes the following components:

[0110] A data reception and caching unit that receives various raw data from a data acquisition module and stores it temporarily;

[0111] A data cleaning unit that performs outlier detection, denoising, and interpolation on the raw data to improve data quality and reliability;

[0112] A data standardization unit that converts data from different sources and scales to a unified standard range for subsequent analysis and comparison;

[0113] A time synchronization unit that aligns the time of data from different sensors to ensure data consistency in the time dimension;

[0114] A data fusion unit that integrates multi-source data after cleaning, standardization, and synchronization to generate a unified data structure;

[0115] An output formatting unit that converts the fused data into a predefined standard format for subsequent module processing and analysis.

[0116] In this embodiment, the data reception and caching unit is the starting point of the entire data preprocessing process, ensuring that raw data can be transmitted to subsequent processing modules in a timely and effective manner. The design of this unit needs to have high throughput and low latency to handle real-time data streams. The caching mechanism not only helps to handle data transmission latency but also provides a certain degree of redundancy to prevent data loss. During the caching process, it is also necessary to consider the data storage format and access speed to support efficient data processing.

[0117] In this embodiment, outlier detection is used to identify and process outliers in the data, which may be caused by sensor failures or external interferences. Denoising removes random noise in the data through filtering algorithms (such as low-pass filters, median filters, etc.) to improve the smoothness of the data. Interpolation processing is used to fill in missing data points to ensure data integrity. These processing steps help to improve data accuracy and lay a good foundation for subsequent analysis and fusion.

[0118] In this embodiment, the data standardization unit converts data from different sources and scales to a unified standard range. Since different sensors may use different units and ranges, it may be difficult to directly compare and analyze this data. The standardization process converts the data to a unified scale and range for effective comparison and analysis. For example, all data is standardized to the range of 0 to 1, or it is converted to a standard normal distribution with a mean of 0 and a standard deviation of 1. This step not only improves data consistency but also simplifies the subsequent processing process.

[0119] In this embodiment, the time synchronization unit aligns the data from different sensors in time to ensure the consistency of the data in the time dimension. Since the data acquisition frequencies of different sensors may be different, directly combining these data may lead to time mismatches. The time synchronization unit aligns the data of all sensors to the same time reference through timestamps or interpolation methods. For example, linear interpolation or spline interpolation can be used to fill in the missing data at different time points, thereby ensuring the consistency of all data points in time. This process is crucial for subsequent data fusion and analysis, ensuring the time continuity and accuracy of the data.

[0120] In this embodiment, the data fusion unit integrates the multi-source data that has been cleaned, standardized, and synchronized to generate a unified data structure. During the fusion process, it is necessary to combine the data from different sensors into a unified format for comprehensive analysis. Multiple methods can be used for data fusion, such as the weighted average method, Kalman filter, Bayesian fusion, etc. These methods generate a more accurate and comprehensive data representation by combining the advantages of each data source. For example, the position information from multiple sensors is weighted and fused to obtain the accurate position of the user. This process not only improves the accuracy of the data but also reduces the impact caused by the errors of a single sensor.

[0121] In this embodiment, the output formatting unit converts the fused data into a predefined standard format for the subsequent processing and analysis of modules. The formatting standards can include the structure of the data, field names, data types, etc. The standardized data format helps to seamlessly interface with other systems or modules, ensuring the compatibility and usability of the data. For example, the fused data is converted into JSON or XML format, or stored according to a specific database table structure. This process simplifies the subsequent processing of the data and improves the overall efficiency and flexibility.

[0122] In summary, the data preprocessing and fusion module ensures that the raw data obtained from the data acquisition module can be provided to the subsequent modules in a high-quality and highly consistent form through a series of refined processing steps. The data reception and caching unit ensures the real-time transmission and storage of the data, the data cleaning unit improves the reliability and accuracy of the data, the data standardization unit unifies the scale and range of the data, the time synchronization unit ensures the time consistency of the data, the data fusion unit generates a unified data structure, and the output formatting unit simplifies the complexity of subsequent processing. Overall, the efficient cooperation of these components improves the data processing ability and provides a solid foundation for intelligent analysis and applications.

[0123] Furthermore, the intelligent analysis module includes the following components:

[0124] Feature extraction unit, which extracts key features from the fused data to provide high-quality input for subsequent analysis;

[0125] Deep learning model unit, which uses a pre-trained neural network model to perform in-depth analysis on the extracted features and identify complex data patterns;

[0126] Time series analysis unit, which uses time series analysis techniques to capture the time dependence and change trends of the data;

[0127] Respiratory pattern recognition unit, which combines the results of deep learning and time series analysis to identify and classify different respiratory patterns;

[0128] Health status assessment unit, which evaluates the user's current health status based on the identified respiratory patterns and other physiological indicators;

[0129] Rehabilitation trend prediction unit, which integrates the user's historical health data and the current assessment results to predict the user's rehabilitation trend and potential risks.

[0130] In this embodiment, the feature extraction unit extracts key features from the fused data to provide high-quality input for subsequent analysis. The main task of this unit is to convert the original data into a more informative feature set, which can better describe the internal patterns and rules of the data. For example, features such as respiratory rate, respiratory depth, heart rate variability, and oxygen fluctuation are extracted from data such as respiratory airflow, heart rate, and blood oxygen saturation. The feature extraction process usually includes signal processing techniques such as Fourier transform and wavelet transform to extract frequency-domain or time-domain features from the original data. High-quality features can not only improve the analysis accuracy of the model but also speed up the subsequent analysis process, ensuring that the system can respond in a timely manner to the user's health changes.

[0131] In this embodiment, the deep learning model unit uses a pre-trained neural network model to perform in-depth analysis on the extracted features and identify complex data patterns. Deep learning models, especially convolutional neural networks (CNNs) and long short-term memory networks (LSTMs), can process a large number of data features and learn complex patterns and rules from them. For example, a CNN can effectively extract local features in the data and perform spatial pattern recognition, while an LSTM is suitable for processing data with time dependence and capturing long-term dependencies. Pre-trained models are usually trained on large-scale datasets and have strong generalization capabilities, which can adapt to new data. The deep learning model unit provides the system with accurate pattern recognition capabilities through in-depth analysis of the extracted features.

[0132] In this embodiment, the timing analysis unit uses time series analysis techniques to capture the time dependence and changing trends of data. Time series analysis is a key technology for processing time-related data, which includes autoregressive models (AR), moving average models (MA), autoregressive moving average models (ARMA), etc. These models can analyze the historical trends, periodic changes, and seasonal fluctuations of data, thereby revealing the inherent time patterns of the data. Timing analysis can help the system understand the time dynamics of user health data. For example, it can identify the changing trends of breathing patterns and the fluctuation patterns of heart rates. This process provides strong support for predicting the user's health status and rehabilitation trends.

[0133] In this embodiment, the task of the breathing pattern recognition unit is to identify the user's breathing patterns, such as normal breathing, shallow breathing, intermittent breathing, etc., from the extracted features and analysis results. Deep learning models can identify different breathing patterns by learning a large amount of labeled data, while timing analysis helps to understand the changes of these patterns over time. The pattern recognition results are crucial for adjusting the training plan and evaluating the user's breathing function. For example, if an abnormal breathing pattern of the user is identified during the training process, the system can immediately adjust the training content to optimize the effect.

[0134] In this embodiment, the health status assessment unit comprehensively considers the user's breathing pattern, heart rate, blood oxygen level, and other relevant physiological indicators to evaluate the user's overall health status. The assessment can be carried out through preset health standards, thresholds, or comparisons based on historical data. For example, it can be determined whether the user has health problems, such as difficulty breathing or cardiovascular abnormalities, according to the breathing pattern and physiological indicators within the normal range. Accurate health status assessment helps users understand their own health status in a timely manner and provides a scientific basis for rehabilitation training.

[0135] In this embodiment, the rehabilitation trend prediction unit analyzes the user's health historical data and current status, and uses prediction models (such as regression models, time series prediction models) to predict future health trends. For example, it can predict the changing trends of the user's breathing function and potential health risks in a future period. The prediction results help users and medical staff formulate personalized rehabilitation plans and preventive measures, and improve the effectiveness and safety of rehabilitation training.

[0136] In summary, through steps such as feature extraction, deep learning, time series analysis, respiratory pattern recognition, health status assessment, and rehabilitation trend prediction, the intelligent analysis module realizes a comprehensive analysis and accurate assessment of the user's health data. The feature extraction unit provides high-quality input data, the deep learning model unit identifies complex data patterns, the time series analysis unit captures time dependencies and trends, the respiratory pattern recognition unit classifies different respiratory patterns, the health status assessment unit provides the current health status assessment, and the rehabilitation trend prediction unit predicts future health trends. The collaborative work of these components provides strong support for personalized rehabilitation training and health management, improving the intelligence level and user experience of the system.

[0137] Furthermore, key features are extracted from the fused data through the following steps:

[0138] Feature screening is performed on the fused multi-source data to accurately extract features highly relevant to respiratory training, while filtering out redundant information irrelevant to respiratory training to optimize the efficiency and accuracy of subsequent analysis. Feature screening is the first step in the feature extraction process, and its main goal is to accurately extract features highly relevant to respiratory training from the fused multi-source data while filtering out redundant information irrelevant to respiratory training. This step is achieved by applying feature selection algorithms (such as mutual information, correlation coefficient analysis, principal component analysis (PCA), etc.) to evaluate the correlation of each feature with the target variables (such as respiratory effect, health status). Through screening, features that have a significant impact on respiratory training can be retained, thus optimizing the efficiency and accuracy of subsequent analysis. For example, features directly related to respiratory rate, respiratory depth, heart rate variability, etc. can be screened out, while features with less association, such as certain environmental parameters or irrelevant physiological indicators, can be removed.

[0139] The fixed time window method is used to segment the data to capture the dynamic change trends within a specific time range. The fixed time window method divides the continuous data stream into several windows of a fixed time length, and the data within each window is used for separate analysis. This method can effectively capture the dynamic change trends in the short term and compare them with the long-term trends. For example, each time window is set to 5 seconds, and the data features within each 5-second window are analyzed. This segmentation helps to identify changes that occur within a specific time period, such as sudden changes in respiratory patterns or fluctuations in heart rate.

[0140] Statistical analysis is performed on the data within each time window to extract key statistical features including mean, variance, maximum value, minimum value, and skewness, comprehensively characterizing the data distribution characteristics.

[0141] The time series data is transformed into the frequency domain through Fourier transform to extract spectral features, effectively identifying the periodic change patterns in respiration and heart rate. Fourier transform decomposes the time series data into sine waves and cosine waves of different frequencies, enabling us to analyze the frequency components of the data. For example, through Fourier transform, the frequency of the respiratory cycle, the variation period of the heart rate, and other periodic characteristics can be discovered. These spectral features are of great significance for identifying regular patterns and abnormal patterns, such as discovering patterns of periodic breathing disorders or irregular heart rates.

[0142] Standardization processing is implemented on the extracted feature set to ensure that all features have a unified dimension. Common standardization methods include: Normalization: Transforming the feature values into the range of 0 to 1; Standardization: Transforming the feature values into a standard normal distribution with a mean of 0 and a standard deviation of 1. Standardization processing helps to eliminate the dimensional differences between features, enabling each feature to have the same weight in model training and improving the training effect and accuracy of the model.

[0143] In summary, the feature extraction process ensures the extraction of key and highly relevant features from the fused data through a series of refined steps. Feature screening helps to optimize the data quality, the fixed time window method can capture dynamic changes, statistical analysis provides a basic description of the data, Fourier transform reveals periodic patterns, and feature standardization ensures data consistency. These steps cooperate with each other, providing high-quality input for subsequent in-depth analysis and model training, improving the efficiency and accuracy of the overall analysis, and laying a solid foundation for the implementation of the intelligent breathing training system.

[0144] Furthermore, the mean μ of the data within the m-th time window m is expressed as: μ m = 1 / n · ∑ i＝1 n X m (i), where n is the number of data points within the window, and X m (i) is the i-th data point within the time window;

[0145] The variance σ of the data within the m-th time window m 2 is expressed as: σ m 2 = 1 / n · ∑ i＝1 n (X m (i) - μ m ) 2 ;

[0146] The maximum value X of the data within the m-th time window max,m is: X max,m = max i＝1,...,n Xm (i);

[0147] The minimum value X of the data within the m-th time window min,m is: X min,m = min i＝1,...,n X m (i);

[0148] The skewness γ of the data within the m-th time window m is expressed as: γ m = 1 / n·∑ i＝1 n (X m (i) - μ m ) / σ m ) 3 , where σ m represents the standard deviation of the data within the m-th time window.

[0149] In summary, the calculation of the mean, variance, maximum value, minimum value, and skewness of the data within the m-th time window provides a comprehensive description of the data. These statistical features reveal the central tendency, degree of fluctuation, extreme values, and symmetry of the distribution of the data. The mean and variance help to understand the basic level and variability of the data, the maximum and minimum values provide information about the range of the data, and the skewness reveals the distribution characteristics of the data. Combining these features, the health status and training effect of the user within a specific time period can be comprehensively evaluated, providing a solid foundation for subsequent data analysis and model training.

[0150] Furthermore, the time series data is transformed into the frequency domain through Fourier transform to extract spectral features, including the following steps:

[0151] Calculate the spectrum of the time series data using the discrete Fourier transform. The specific formula is: X[k] = ∑ f＝0 N-1 x[f]·e -j2kfπ / N , where X[k] is the k-th frequency component, representing the complex value of the signal at frequency k; x[f] is the value of the time series data at the f-th time point; e -j2kfπ / N is the complex exponential function, representing the phase information at frequency k; j is the imaginary unit, satisfying j 2 = -1; N represents the total number of sampling points in the time series data, that is, the total number of data points; f represents the f-th sampling point in the time series, that is, the position index in the current time domain signal, and f takes integer values from 0 to N - 1; k represents the k-th frequency component in the frequency domain, which corresponds to the frequency index after Fourier transform, and the value range of k is 0 to N - 1, representing the components from the lowest frequency to the highest frequency;

[0152] Extract the amplitude spectrum and phase spectrum from the frequency spectrum as frequency spectrum features: A[k] = |X[k]|, φ[k] = arg(X[k]), where A[k] represents the amplitude of the k-th frequency component, indicating the intensity of the signal at that frequency; |X[k]| represents the modulus of X[k], that is, the amplitude of the signal at frequency k; φ[k] represents the phase angle of the k-th frequency component, indicating the phase information of the signal at that frequency; arg(X[k]) represents the phase angle of X[k].

[0153] In summary, by transforming time series data to the frequency domain through Fourier transform, the frequency spectrum features of the signal can be effectively extracted. The discrete Fourier transform provides the complex values of the signal at different frequencies, and the amplitude spectrum and phase spectrum respectively represent the intensity and phase information of the frequency components. These frequency spectrum features are of great significance for identifying periodic changes in the signal, analyzing respiratory patterns and heart rate changes, etc. The amplitude spectrum helps to identify the intensity of the main frequency components, while the phase spectrum provides the phase information of the frequency components. Combining these features can provide a deeper understanding of the frequency characteristics of the signal and provide key data for further analysis and applications.

[0154] Furthermore, combining deep learning and the results of time series analysis, different respiratory patterns are identified and classified, including the following steps:

[0155] Integrate the output results of the deep learning model and time series analysis into a unified high-dimensional feature set, and achieve seamless integration of heterogeneous data through feature splicing technology. This process includes: Output of the deep learning model: The deep learning model extracts high-level abstract features by processing the original data, and these features include complex non-linear patterns and high-level representations; Results of time series analysis: Time series analysis provides the time dependence and trend information of the data, including periodic features, fluctuation patterns, etc. Feature splicing technology is used to achieve seamless integration of these heterogeneous data. Through the splicing technology, the features extracted by deep learning are combined with the features obtained from time series analysis to form a high-dimensional feature set. These high-dimensional feature sets contain rich signal information and can more comprehensively describe various dimensions of the respiratory pattern. For example, the deep features extracted from a convolutional neural network (CNN) are spliced with the time series features obtained based on an autoregressive model (AR) to generate a comprehensive feature vector.

[0156] Based on the high-dimensional feature set, a dedicated breathing pattern classifier is constructed to accurately identify and classify various breathing patterns. This step includes: Model selection: Select an appropriate classification model according to the characteristics of the data and the task requirements. Common classifiers include Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting Tree (GBT), and Deep Neural Network (DNN), etc.; Training: Use the labeled training data to train the classifier. The training process includes the optimization and learning of model parameters, with the goal of enabling the classifier to accurately identify and classify various breathing patterns; Validation: Evaluate the performance of the model through techniques such as cross-validation to ensure that the classifier has good generalization ability and avoid overfitting phenomena.

[0157] Compare and verify the classification results with the actual observed data to comprehensively evaluate the performance metrics of the classifier in the recognition of different breathing patterns. This process includes: Performance metric evaluation: Use metrics such as accuracy, precision, recall, and F1-score to comprehensively evaluate the performance of the classifier in the recognition of different breathing patterns. For example, accuracy measures the overall recognition accuracy of the classifier, while the F1-score combines precision and recall; Comparison with actual observed data: Compare the prediction results of the classifier with the actually observed breathing patterns to verify the actual application effect of the classifier. Through comparison, the advantages and disadvantages of the classifier can be identified.

[0158] Based on the evaluation results, implement a series of optimization measures, including parameter fine-tuning, feature selection improvement, and training dataset expansion. Through a continuous iterative optimization process, continuously improve the overall performance of the classifier. Optimize the performance of the model by adjusting the hyperparameters of the classifier (such as learning rate, regularization coefficient, etc.). Common parameter optimization methods include grid search and Bayesian optimization; According to the evaluation of feature importance, select the features that are most useful for the classification task, and remove redundant or irrelevant features. Feature selection can improve the efficiency and accuracy of the model; By increasing the amount of training data (such as collecting more samples or performing data augmentation), improve the generalization ability and recognition accuracy of the classifier. Data expansion helps to improve the model's ability to recognize various breathing patterns. Through continuous iterative optimization, periodically update and improve the classifier. This includes continuously collecting new data, retraining the model, adjusting features and parameters, so as to continuously improve the overall performance of the classifier.

[0159] In summary, the respiratory pattern recognition and classification by integrating deep learning and time series analysis results involve multiple key steps, such as feature integration, classifier construction, result verification, and optimization. By integrating the output features of the deep learning model and time series analysis, a comprehensive high-dimensional feature set can be created, providing rich data support for the classifier. Constructing a dedicated classifier and verifying its effectiveness through actual data can ensure the accuracy and practicality of the classifier. Through optimization measures, including parameter fine-tuning, feature selection, and dataset augmentation, the performance of the classifier can be further improved to achieve efficient and accurate respiratory pattern recognition. These steps cooperate with each other to form a complete respiratory pattern recognition process, providing strong technical support for related applications.

[0160] Furthermore, based on the recognized respiratory pattern and other physiological indicators, the current health status of the user is evaluated, including the following steps:

[0161] Integrate the recognized respiratory pattern with heart rate, blood oxygen saturation, and body temperature indicators to construct a multi-dimensional comprehensive health dataset. This process includes: Data integration: Collect and organize data from different sources, such as respiratory patterns (identified by the intelligent analysis module), heart rate (measured by a photoplethysmography sensor), blood oxygen saturation (also measured by a photoplethysmography sensor), body temperature (measured by a body temperature sensor), etc., and integrate them into a multi-dimensional health dataset; Data standardization: Standardize the indicators of different data sources to ensure that they can be compared and analyzed under the same dimension. This includes normalizing and scaling the data so that each indicator in the dataset can be seamlessly integrated.

[0162] Based on medical literature and expert consensus, a set of health standard systems are established, including the normal respiratory pattern range and physiological indicator reference values, as the benchmark for evaluating the user's health status. This process includes: Normal respiratory pattern range: Determine the range of normal respiratory frequency and depth. For example, the normal respiratory frequency of adults is usually between 12 and 20 breaths per minute; Physiological indicator reference values: Determine the normal reference value ranges of heart rate, blood oxygen saturation, and body temperature. For example, the normal heart rate is 60 to 100 beats per minute, the blood oxygen saturation should usually be between 95% and 100%, and the normal body temperature is 36.1°C to 37.2°C.

[0163] Conduct multi-dimensional comparative analysis of the user's comprehensive health data set with the set health standards to accurately evaluate the user's overall health status. The comparative analysis of the user's comprehensive health data set with the set health standards includes: Comparative analysis: Compare item by item the indicators such as the user's breathing pattern, heart rate, blood oxygen saturation, body temperature, etc. with the normal range standards. For example, if the user's heart rate is higher than 100 beats per minute, it may indicate a problem of too fast heart rate; Comprehensive evaluation: Comprehensively evaluate the user's overall health status by combining multi-dimensional data. For example, abnormalities in heart rate and breathing pattern may indicate problems with cardiopulmonary function.

[0164] Adopt anomaly detection algorithms to monitor the deviation degree of the user's physiological indicators in real time, mark the indicators beyond the normal range as potential health risks, and conduct in-depth cause analysis and impact assessment. Use statistical methods or machine learning models to detect outliers in the data. For example, use Z-score-based or model-based anomaly detection methods to identify indicators that deviate from the normal range. Mark the indicators beyond the normal range as potential health risks. For example, a blood oxygen saturation lower than 95% may indicate breathing problems. Conduct in-depth analysis of the detected anomalies and evaluate their possible impact on the user's health. For example, a persistent high heart rate may be related to overexertion, stress, or potential heart problems.

[0165] Based on the comprehensive results of health status analysis and anomaly detection, adopt a weighted scoring mechanism to assign a quantitative score to the user's health status. Set weighted coefficients according to the importance and anomaly degree of each indicator, and assign a comprehensive score to the user's health status. For example, an anomaly in blood oxygen saturation may receive a higher weight because it directly affects respiratory function. Calculate a comprehensive health score by integrating the anomaly degree and health impact of all indicators. For example, use the weighted average method to summarize the scores of each indicator.

[0166] According to the health status score and individual characteristics, use decision support algorithms to generate personalized health advice and intervention plans to provide accurate health management guidance for users.

[0167] In summary, by integrating the identified breathing pattern with other physiological indicators to construct a comprehensive health data set and establishing a health assessment system based on medical standards, the user's health status can be comprehensively evaluated. Using multi-dimensional comparative analysis, anomaly detection, and impact assessment, potential health risks can be accurately identified. By assigning a quantitative score to the user's health status through a weighted scoring mechanism and generating personalized health advice and intervention plans based on this, accurate health management guidance can be provided. These steps form a systematic health assessment process, which helps to improve the effectiveness and accuracy of user health management.

[0168] Furthermore, the deviation degree of physiological indicators is represented by D 偏离It is expressed, and the calculation formula is: D 偏离 =(x - μ 正常 ) T S -1 (x - μ 正常 ), where x is the vector of physiological index data monitored in real time at the current time point; μ 正常 is the mean vector of physiological indexes in the normal state; S is the covariance matrix of the physiological index data monitored in real time at the current time point, which is used to measure the correlation and variability between various indexes; T represents the transpose operation; (x - μ 正常 ) T refers to the transpose of the difference vector between x and μ 正常 ; S -1 is the inverse matrix of the covariance matrix S.

[0169] In summary, by calculating D 偏离 , the abnormal degree of the user's physiological indexes can be evaluated in real time, helping to identify potential health risks. For example, when D 偏离 exceeds the set threshold, it can be marked as a health risk and corresponding intervention measures can be taken. This method helps to achieve more accurate abnormal detection and health status assessment in the health management system.

[0170] Furthermore, the training plan generation module includes the following components:

[0171] User profile management unit, which stores and updates the user's long-term health data, training history and personal preferences, providing background information for plan generation;

[0172] Training goal setting unit, which sets and manages short-term and long-term training goals according to medical advice and user wishes;

[0173] Real-time status input unit, which receives and processes the real-time user status data from the intelligent analysis module to ensure that the plan matches the current situation;

[0174] Training plan generator, which comprehensively considers the user profile, training goals and real-time status to generate a preliminary personalized training plan;

[0175] Plan adaptability adjustment unit, which dynamically adjusts and optimizes the training plan according to the user's real-time feedback and training progress to ensure its continuous effectiveness;

[0176] Training instruction output unit, which converts the generated training plan into specific and executable training instructions and transmits them to the real-time feedback and interaction module.

[0177] In this embodiment, the user profile management unit uses database technology to systematically record the user's health information, including medical history, past training effects, device usage habits, etc. By analyzing this data, personalized training suggestions can be provided for each user. The storage of data not only ensures the security of information but also enables rapid retrieval and update when needed, so as to reflect the user's latest health status and training situation in real time. This ability to track over a long period can identify and predict the changing trends of the user's health status, thereby providing a more scientific basis for subsequent training programs and ensuring the effectiveness and sustainability of the rehabilitation process.

[0178] In this embodiment, the training goal setting unit relies on the professional opinions of medical staff and the user's own feedback, combines medical research and training effect data, and sets specific rehabilitation goals. For example, for certain diseases, short-term goals may include increasing lung capacity, while long-term goals may be to restore near-normal breathing function. Goal setting not only needs to consider the user's current health status but also combines the user's rehabilitation progress to ensure that the goals are practically feasible and challenging. By setting clear goals, this unit can provide a clear direction for the training process and also provide a reference basis for subsequent program adjustment and effect evaluation, thereby enhancing the pertinence and effectiveness of training.

[0179] In this embodiment, the real-time status input unit, through seamless integration with sensors and data acquisition modules, can monitor the user's physiological status in real time, such as respiratory rate, respiratory depth, and oxygen saturation. Through high-frequency data acquisition and processing, the user's health changes and training effects can be immediately understood. This data is then transmitted to the training program generation module to ensure that the generated training program can timely reflect the user's actual status and thus make timely adjustments. The efficiency and accuracy of the real-time status input unit directly affect the dynamic adjustment ability of the training program and are key factors in ensuring the matching of personalized training programs with the user's actual needs.

[0180] In this embodiment, the training program generator uses advanced algorithms and models to match the collected user data with the set training goals and automatically generate a training plan that meets the user's actual needs. The generator usually includes multiple algorithm models, such as optimization algorithms and machine learning models, to ensure that the training program meets both medical standards and the user's personalized needs. Through in-depth analysis of user data, the generator can formulate a scientific and reasonable training plan, thereby improving the efficiency and effect of rehabilitation training.

[0181] In this embodiment, the program adaptability adjustment unit depends on the data provided by the real-time status input unit. By analyzing the user's training feedback, it timely modifies the training program. The adjustment content includes changing the training intensity, frequency or content to adapt to the user's rehabilitation progress and physical state changes. The implementation of this function ensures that the training program can flexibly adapt to the changing needs of the user, improving the effectiveness and safety of training. The dynamic optimization ability of the program adaptability adjustment unit makes the training process more personalized and intelligent, and can continuously improve the user's rehabilitation effect.

[0182] In this embodiment, the training instruction output unit converts the details of the training program, such as training actions, frequency, duration, etc., into specific instructions that the user can understand and execute. Through the interface for interacting with the user, the training instructions can be conveyed to the user in real time and guide them to perform correct training. This unit not only needs to ensure the accuracy of the instructions, but also needs to pay attention to the user's execution situation in order to make timely adjustments and optimizations. The effectiveness of the training instruction output unit directly affects the implementation effect of the training program and is a key link in realizing the personalized training plan.

[0183] In summary, through the coordinated work of the user profile management unit, training goal setting unit, real-time status input unit, training program generator, program adaptability adjustment unit and training instruction output unit, the training program generation module realizes the generation of a personalized and dynamically adjustable training program. These components each play an important role, jointly ensuring the scientificity, effectiveness and operability of the training program. By continuously tracking the user's health status and feedback, the system can optimize the training program in real time, improve the rehabilitation effect, and provide a more accurate and effective rehabilitation training experience for the user.

[0184] Furthermore, comprehensively considering the user profile, training goals and real-time status, a preliminary personalized training program is generated, including the following steps:

[0185] Through multi-dimensional user profile analysis, a comprehensive assessment is made of the user's long-term health data, historical training records, personal health status, past medical history, motor ability and preferences. This step includes an in-depth assessment of the user's long-term health data, historical training records, personal health status, past medical history, motor ability and personal preferences. By integrating this information, a comprehensive user health profile can be established. By analyzing these data, the user's health trends, potential health problems and the current status of motor ability can be identified. This multi-dimensional assessment provides important background information for the formulation of the training program, helping to ensure that the generated program can truly meet the actual needs and expectations of the user.

[0186] Using a goal-setting algorithm, combined with the user's health goals and needs, to formulate precise training goals that match their health status and training ability. After understanding the user's basic health profile, the goal-setting algorithm will be used to formulate precise training goals. These goals not only need to meet the user's health needs but also consider the user's current training ability and health status. The algorithm will match the user's health goals (such as improving lung capacity, enhancing physical strength, etc.) with the user's actual ability to ensure that the set goals are both challenging and within a reasonable range. For example, for a user who hopes to improve their respiratory function, the goals may include gradually increasing the daily duration of respiratory training or increasing the training intensity. By precisely setting goals, the system can provide a clear training direction and promote the user's rehabilitation progress.

[0187] Using the physiologic data and training progress information collected in real time to dynamically evaluate the user's current state. This step is a key step in the training plan generation process. By monitoring the user's physiologic state in real time, such as indicators like heart rate and respiratory rate, to judge the user's current health status and training effect. Comparing with the user's goals, the effectiveness of the training and the user's adaptation can be evaluated. If it is found that the user's state does not match the expectation, these changes can be captured in real time, thereby adjusting the training plan. The dynamic evaluation of the real-time state ensures the timeliness and effectiveness of the training plan, enabling the training plan to adapt to the user's actual situation.

[0188] Comprehensively considering the user profile, training goals, and real-time state data to generate a preliminary personalized training plan. The personalized training plan will combine the user's historical data and current state to provide specific training content, frequency, intensity, and duration. For example, the plan may include specific respiratory training, physical training, or other relevant rehabilitation activities. The generated plan at this stage provides the user with a clear training plan, aiming to improve the rehabilitation effect through scientific methods.

[0189] Adopting a feedback optimization mechanism, according to the user's real-time feedback and training effect data, automatically adjusting the training intensity, content, and cycle to achieve the dynamic optimization of the training plan. The user's real-time feedback and training effect data will be used to evaluate the actual effect of the training plan. According to the user's feedback, the training intensity, content, and cycle can be automatically adjusted. For example, if the user reports that the training intensity is too high, the training intensity can be reduced or the training frequency can be adjusted. This optimization mechanism ensures the flexibility and adaptability of the training plan, enabling the plan to be continuously improved to achieve the best effect.

[0190] The optimized personalized training plan is presented to the user through a visual interface, along with detailed implementation guidelines and support services. The design of the visual interface aims to enable the user to intuitively understand and execute the training plan. At the same time, support services are also provided to help the user solve problems encountered during implementation. Through these support services, the user can obtain additional guidance and assistance to ensure the smooth implementation of the training plan.

[0191] In summary, generating a preliminary personalized training plan is a multi-step process involving multiple links such as user profile analysis, goal setting, real-time status assessment, plan generation, feedback optimization, and final presentation. Each step ensures the scientific and effectiveness of the training plan while taking into account the user's personalized needs and actual situation. Through comprehensive data analysis and dynamic adjustment mechanisms, this process not only improves the pertinence and effectiveness of training but also provides the user with an intuitive and operable training plan and support, further enhancing the overall effect of rehabilitation training.

[0192] Furthermore, the real-time feedback and interaction module includes the following components:

[0193] The training instruction receiving unit receives specific instructions from the training plan generation module, providing the basic content for feedback generation.

[0194] The multi-modal feedback generator generates feedback content suitable for different sensory channels based on the training instructions and the user's real-time status.

[0195] The visual feedback unit displays visual information including training guidance, progress bars, and performance statistics through a graphical interface.

[0196] The auditory feedback unit provides real-time training guidance and encouraging feedback using voice prompts and sound effects.

[0197] The tactile feedback unit provides training rhythm guidance and completion prompts through tactile signals such as vibration.

[0198] The user input processing unit receives and parses the user's operations and voice instructions to achieve two-way interaction.

[0199] In this embodiment, the training instruction receiving unit is responsible for receiving specific training instructions from the training plan generation module. These instructions include training content, frequency, intensity, and other relevant details, which are the basis for generating feedback content. The main function of the receiving unit is to ensure the accurate and error-free transmission of training instructions and provide basic data for other feedback components. The efficiency of this unit directly affects the timeliness and accuracy of subsequent feedback. If the instruction reception is inaccurate or delayed, it may cause the user to not receive the correct training guidance, thus affecting the training effect.

[0200] In this embodiment, the multimodal feedback generator is the core of the real-time feedback and interaction module. It generates feedback content suitable for different sensory channels based on training instructions and the user's real-time status. This unit generates various forms of feedback by comprehensively analyzing the user's physiological data and training progress information to ensure that the user can receive effective guidance in different situations. The multimodal design of the generator includes visual, auditory, and tactile feedback, which can cover the different sensory needs of the user and improve the acceptability and effectiveness of the feedback. For example, if the user performs poorly at a certain stage, the system can give hints and encouragement through different feedback forms simultaneously.

[0201] In this embodiment, the visual feedback unit displays visual information such as training guidance, progress bars, and performance statistics through a graphical interface. This information helps the user intuitively understand their training status and progress. For example, the progress bar can show the completion status of the training, and the performance statistics can display the user's performance at different training stages. The design of the graphical interface should be simple and clear, making it easy for the user to quickly understand and operate. Good visual feedback can enhance the user's sense of participation and self-management ability, promoting the achievement of training goals.

[0202] In this embodiment, the auditory feedback unit provides real-time training guidance and encouraging feedback using voice prompts and sound effects. The voice prompts can convey training instructions, adjustment suggestions, and progress updates in real time, while the sound effects can provide immediate feedback and motivation during the training process. The design of the auditory feedback needs to consider the user's auditory perception and environmental noise to ensure that the prompts and sound effects can be clearly conveyed without disturbing the user. Through reasonable sound feedback, the user can receive continuous guidance and encouragement during the training process, thereby enhancing the effectiveness and motivation of the training.

[0203] In this embodiment, the tactile feedback unit provides training rhythm guidance and completion prompts through tactile signals such as vibration. This feedback method can use tactile signals to prompt the training rhythm or remind the user to complete a certain stage during the training. For example, when the user needs to perform a certain training at a specific time interval, the tactile feedback can guide the user to execute according to the rhythm through vibration signals. The advantage of tactile feedback is that it can still provide effective instructions and prompts when the user cannot directly see or hear other feedback, improving the accuracy and timeliness of the training.

[0204] In this embodiment, the user input processing unit can receive the user's operation inputs (such as touch screen operations, button clicks) and voice commands, parse their intentions, and feedback corresponding information. Through two-way interaction, the user can make personalized adjustments to the training process, feedback problems, or request help. This interaction ability not only improves the user experience but also can more flexibly meet the user's needs and preferences. The efficiency and accuracy of the user input processing unit directly affect the system's response speed and interaction quality.

[0205] In summary, through the collaborative work of the training instruction receiving unit, multi-modal feedback generator, visual feedback unit, auditory feedback unit, tactile feedback unit, and user input processing unit, the real-time feedback and interaction module provides comprehensive and personalized training guidance and feedback. These components ensure that users can receive timely, accurate guidance and motivation during the training process through multi-sensory feedback forms. Through two-way interaction, users can not only obtain real-time feedback but also make personalized adjustments to the training process, improving the efficiency and effectiveness of training. Overall, the design and implementation of the real-time feedback and interaction module greatly enhance the user's sense of participation and training effect, providing strong support for achieving effective rehabilitation training.

[0206] Furthermore, the effect evaluation and optimization module includes the following components:

[0207] Long-term data recording unit, continuously collecting and storing the user's training data, physiological indicators, and health status, providing comprehensive historical data support for subsequent analysis;

[0208] Effect evaluation engine, based on the recorded long-term data, comprehensively analyzing the user's rehabilitation progress and training effect, generating a quantitative evaluation report;

[0209] Health trend analysis unit, using time series analysis technology to deeply analyze the long-term changes in the user's health status, identifying key health trends and patterns;

[0210] Reinforcement learning optimizer, modeling the training process as a reinforcement learning problem, continuously optimizing the training strategy based on the evaluation results and health trends;

[0211] Policy update unit, receiving the optimized policy and converting it into specific training plan adjustment suggestions to achieve dynamic update of the training strategy;

[0212] Personalized parameter adjustment unit, fine-tuning the optimized training strategy according to the user's individual characteristics and feedback to ensure its adaptability and personalization.

[0213] In this embodiment, the core function of the long-term data recording unit is to provide comprehensive historical data support for subsequent analysis and optimization. It integrates various data collection devices and sensors to record the user's training process and related health data in real time, including heart rate, respiratory rate, exercise volume, sleep quality, etc. The systematic storage and management of long-term data not only help establish a complete health record for the user but also provide a solid foundation for subsequent effect evaluation and trend analysis. The accuracy and integrity of the data are crucial for the scientific nature of the effect evaluation. Therefore, the design of the long-term data recording unit needs to ensure high-frequency and high-accuracy data collection to support subsequent in-depth analysis.

[0214] In this embodiment, the effect evaluation engine uses data analysis and statistical models to compare the user's current state with the predetermined goals and evaluate the effect and progress of the training. Through detailed analysis of the data, the effect evaluation engine can generate a detailed evaluation report, including quantitative indicators of the training effect, trends in the rehabilitation progress, and improvement suggestions. The evaluation report not only provides clear feedback to the user on the training effectiveness but also provides an important basis for system optimization, thus supporting more scientific adjustment of the training plan.

[0215] In this embodiment, by analyzing the trends and patterns in the user's long-term data, the health trend analysis unit can identify key health trends, such as the improvement or deterioration of health indicators, the speed of rehabilitation progress, etc. Time series analysis techniques help discover potential regularities in the data and provide a basis for optimizing the training plan. For example, if the analysis finds that a certain training pattern significantly improves a specific health indicator of the user, this information can be used to adjust the training strategy to enhance the effect. The insights of the health trend analysis unit are crucial for the scientific management of the long-term rehabilitation process and can help the system promptly identify and address potential problems.

[0216] In this embodiment, the reinforcement learning optimizer uses reinforcement learning algorithms to automatically adjust the training strategy according to the training feedback and the user's health data to maximize the training effect. By simulating different training scenarios and strategies, the optimizer can find the optimal training plan to meet the dynamic needs and health changes of the user. The use of the reinforcement learning optimizer makes the adjustment of the training strategy more intelligent and adaptive, and it can continuously improve the training plan based on real-time feedback to enhance the rehabilitation effect.

[0217] In this embodiment, the policy update unit applies the strategy generated by the optimizer to the actual training plan and adjusts the training content, intensity, and cycle according to the user's feedback and health data. The goal of the policy update unit is to ensure the timely update and optimization of the training plan to better meet the user's rehabilitation needs. Through continuous policy updates, the system can maintain the effectiveness and pertinence of the training and ensure that the user can continuously obtain the optimal training effect.

[0218] In this embodiment, the personalized parameter adjustment unit considers the user's specific needs, health status, and training feedback to refine the optimization strategy. By deeply understanding the user's characteristics, the personalized parameter adjustment unit can provide training suggestions that meet individual needs and make the training plan more tailored to the user's actual situation. Personalized adjustment not only improves the user's training experience but also effectively enhances the training effect and rehabilitation progress.

[0219] In summary, through the collaborative work of the long-term data recording unit, the effect evaluation engine, the health trend analysis unit, the reinforcement learning optimizer, the policy update unit, and the personalized parameter adjustment unit, the effect evaluation and optimization module realizes the scientific evaluation of the training effect and policy optimization. The role of each component in the module is crucial, jointly ensuring the continuous improvement and personalized adjustment of the training plan. Through comprehensive data analysis and intelligent optimization, this module not only improves the pertinence and effectiveness of training, but also enhances the user's sense of participation and satisfaction, providing strong support for the rehabilitation process.

[0220] Furthermore, based on the recorded long-term data, comprehensively analyze the user's rehabilitation progress and training effect, and generate a quantitative evaluation report, including the following steps:

[0221] Extract the user's complete training history, the change trajectory of health indicators, and the rehabilitation progress record from the historical database. This step involves extracting all relevant records of the user from the stored long-term data, including training logs, physiological data, health examination results, etc. This process requires efficient data retrieval techniques to ensure that the full set of user data can be obtained quickly. The extracted data includes the user's training frequency, intensity, duration, changes in health indicators such as heart rate and respiratory rate, and other rehabilitation-related information. These data provide the basis for subsequent analysis, ensuring that the evaluation report can reflect the user's true rehabilitation situation and training effect.

[0222] Purify and standardize the extracted recorded data through a series of data preprocessing techniques to improve data quality and analysis applicability. This step mainly includes data cleaning (such as handling missing values and outliers), data transformation (such as standardizing the units and formats of different data sources), data integration (such as integrating data from different sources), etc. The purpose of data purification is to eliminate noise and errors in the data, ensuring the accuracy and reliability of the analysis results. Standardization processing is to convert the data into a unified format and scale for subsequent analysis and comparison. The quality of this step directly affects the effectiveness of subsequent analysis, so it is necessary to ensure the accuracy and comprehensiveness of the processing process.

[0223] Using multi - dimensional data analysis techniques, deeply mine the processed data to identify the key nodes of the user's rehabilitation progress and the quantitative indicators of training effects, and reveal the long - term change trends of health status and the effects of training interventions. Deeply mining the processed data using multi - dimensional data analysis techniques is the core step in generating a quantitative assessment report. By applying statistical analysis, data mining, and machine learning techniques, the key nodes of the user's rehabilitation progress and the quantitative indicators of training effects can be identified. These techniques include regression analysis, clustering analysis, association rule mining, etc., which are used to reveal the long - term change trends of health status and the effects of training interventions. For example, it is identified that certain training phases have significantly improved the user's health indicators, or it is found that the user's rehabilitation progress is slow during a specific period. The results of multi - dimensional data analysis can reveal the key factors in the user's rehabilitation process, providing valuable insights for further optimizing the training plan.

[0224] Through feature engineering and index screening algorithms, extract key indicators from the comprehensive analysis results, including the main parameters of rehabilitation progress and the quantitative indicators of training effects. This step involves selecting and refining various data features in the analysis results to ensure that the information in the report can effectively reflect the user's rehabilitation progress and training effects. Feature engineering involves constructing new features (such as composite indicators) and selecting the most representative features to highlight important health and training indicators. For example, the main parameters extracted from health data may include the improved vital capacity, recovery time after training, etc. Through precise index screening, the generated assessment report can clearly display the user's rehabilitation situation and training effectiveness, avoiding information redundancy and complexity.

[0225] Based on the extracted key indicators and comprehensive analysis results, generate a comprehensive quantitative assessment report, including a summary of the user's rehabilitation progress, an assessment of training effects, data charts, and trend charts, to clearly display the user's overall health changes and training effectiveness. The process of generating the assessment report includes transforming the analysis results into text descriptions, graphical displays, and data tables that are easy to understand and interpret. For example, the summary of rehabilitation progress may include the user's performance in different training phases, and the assessment of training effects may display specific effect indicators and improvement suggestions. Data charts and trend charts help users intuitively understand health changes and training effects, thus providing a clear perspective to evaluate the effectiveness of training. The assessment report is not only used to feedback the user's training progress but also provides a guiding basis for future training plans.

[0226] In summary, the process of generating a quantitative evaluation report provides a comprehensive and accurate assessment of the user's rehabilitation and training effects through steps such as data extraction, preprocessing, multi-dimensional analysis, feature engineering, and report generation. The implementation of each step aims to ensure the accuracy of the data and the depth of the analysis, thereby generating a detailed and useful report. Through these steps, the health changes and training achievements of the user can be clearly demonstrated, providing solid data support for further optimizing the training plan and providing personalized rehabilitation suggestions. Such a quantitative evaluation report not only improves the user's understanding and participation but also provides a scientific basis for the continuous improvement and effectiveness enhancement of the system.

[0227] As Figure 2 shown, another object of the present application is to provide a portable breathing function rehabilitation training method, including the following steps:

[0228] Step 1: Real-time collect the user's physiological parameters, posture information, and environmental parameters;

[0229] Step 2: Clean, standardize, and synchronize the collected data in terms of time, and fuse them into a unified format;

[0230] Step 3: Extract the key features of the user's breathing function from the fused data, identify the breathing pattern, and conduct real-time health status assessment;

[0231] Step 4: Generate a preliminary personalized training plan according to the user's personal profile, training goals, and real-time health status;

[0232] Step 5: Provide instant training guidance for the user through multi-modal feedback and feedback the current progress;

[0233] Step 6: Continuously record the user's long-term training data and generate a quantitative training effect evaluation report;

[0234] Step 7: Update and fine-tune the training strategy according to the evaluation results to achieve the adaptive optimization and personalized iteration of the training plan.

[0235] In Step 1, the portable breathing function rehabilitation device needs to real-time collect the user's physiological parameters (such as heart rate, respiratory rate, oxygen saturation, etc.), posture information (such as body position, postural stability, etc.), and environmental parameters (such as temperature, humidity, air quality, etc.). These data are collected through built-in sensors and monitoring devices and transmitted to the processing system. Real-time collection is the basis for ensuring that the training plan can accurately reflect the user's current state and environmental changes and is crucial for achieving effective rehabilitation training.

[0236] In Step 2, data cleaning includes handling missing values, outliers, and noise to ensure data integrity and reliability. Standardization is to transform data into a unified format and scale, facilitating the comparison and integration of data from different sources. Time synchronization ensures that the timestamps of data are consistent, enabling different types of data to be accurately aligned. Through these preprocessing steps, the data will be transformed into a unified format for subsequent analysis and processing.

[0237] In Step 3, after the data processing is completed, key features of the user's respiratory function will be extracted from the fused data. This includes identifying the user's breathing pattern, analyzing the changes in breathing frequency, and identifying abnormalities in the breathing cycle, etc. Through advanced analysis techniques (such as signal processing, feature extraction algorithms), features of great significance for respiratory function rehabilitation can be extracted from complex datasets. This process is a prerequisite for real-time health status assessment and generating personalized training programs.

[0238] In Step 4, based on the user's personal profile (including health history, training records, etc.), set training goals (such as improving vital capacity, enhancing respiratory endurance, etc.), and real-time health status (data obtained from feature extraction), a preliminary personalized training program is generated. The program includes specific training content, intensity, frequency, and duration, aiming to be adjusted according to the user's specific needs. The generation of the personalized training program takes into account the user's health condition and rehabilitation goals to ensure the pertinence and effectiveness of the training.

[0239] In Step 5, to ensure that users can effectively execute the training program, immediate training guidance and progress feedback are provided through multimodal feedback (including visual, auditory, and tactile feedback). For example, visual feedback may display the training progress and guidance through a graphical interface; auditory feedback may include voice prompts and encouragement; tactile feedback may prompt the user's training rhythm through vibration signals. The design of multimodal feedback aims to enhance the user's sense of participation and training effect, ensuring that users can accurately follow the training program.

[0240] In Step 6, during the training process, the user's long-term training data will be continuously recorded, including various aspects of the training and changes in health indicators. These data will be used to generate a quantitative training effect evaluation report. The evaluation report includes a detailed analysis of the training effect, changes in the user's health condition, and quantitative indicators of the rehabilitation progress, etc. The quantitative report provides users with an intuitive understanding of the training effect and a basis for further training adjustment.

[0241] In step 7, based on the results in the quantitative evaluation report, the training strategy will be updated and fine-tuned. This includes adjusting the content, intensity, frequency, etc. of the training to adapt to the user's rehabilitation progress and new needs. Adaptive optimization and personalized iteration ensure the dynamic adjustment of the training plan, enabling it to be continuously improved and optimized to achieve the best rehabilitation effect. The system enhances the effectiveness of the training plan and user satisfaction through continuous feedback loops.

[0242] In summary, the portable respiratory function rehabilitation training method realizes the scientific training and rehabilitation of the user's respiratory function through steps such as real-time data collection, data preprocessing, feature extraction, personalized plan generation, multi-modal feedback, quantitative evaluation, and adaptive optimization. The implementation of each step aims to ensure the accuracy and effectiveness of the training plan while enhancing the user experience and sense of participation. Through these systematic steps, the portable device can provide a continuously improving training plan to support the user in continuously optimizing their respiratory function during the rehabilitation process and achieving better health effects.

[0243] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A portable respiratory function rehabilitation training device, characterized in that: Includes the following modules: The data acquisition module integrates a variety of micro sensors to collect the user's respiratory airflow, heart rate, blood oxygen saturation and environmental parameters in real time; The data preprocessing and fusion module cleans, standardizes and time-synchronizes the collected raw data and fuses them into a unified data format; Intelligent analysis module, responsible for real-time analysis of fused data, identifying breathing patterns, assessing current health status, and predicting recovery trends; The training program generation module dynamically generates and adjusts personalized breathing training programs based on intelligent analysis results and preset training goals, ensuring that the training content always matches the user's real-time status and needs; Real-time feedback and interaction module, providing users with instant training guidance and progress feedback through visual, auditory and tactile means; The effect evaluation and optimization module continuously records and analyzes the user's long-term training data, evaluates the rehabilitation effect, and uses reinforcement learning algorithms to continuously optimize the training strategy to achieve personalized long-term health management; Portable hardware integration module integrates all modules into a small, low-power portable device to enable training anytime and anywhere.

2. A portable respiratory function rehabilitation training device according to claim 1, characterized in that: The data acquisition module includes the following components: Airflow sensor unit, which measures the user's breathing airflow rate and breathing frequency in real time, providing basic data for analyzing breathing patterns and intensity; A photoplethysmography sensor unit that simultaneously measures the user’s heart rate and blood oxygen saturation using photoelectric technology; Environmental sensor unit, which monitors the temperature, humidity and air quality parameters of the training environment in real time to ensure the comfort and safety of the training environment; The posture sensor unit uses an accelerometer and a gyroscope to collect the user's posture information in real time to ensure that breathing training is performed in the correct state; The body temperature sensor unit is used to detect the user's skin surface temperature and evaluate the user's body temperature changes during training in real time.

3. A portable respiratory function rehabilitation training device according to claim 1, characterized in that: The data preprocessing and fusion module includes the following components: The data receiving and caching unit receives various raw data from the data acquisition module and stores them temporarily; Data cleaning unit, which performs outlier detection, denoising and interpolation processing on raw data to improve data quality and reliability; Data standardization unit, which converts data from different sources and scales into a unified standard range to facilitate subsequent analysis and comparison; The time synchronization unit aligns the data from different sensors to ensure the consistency of the data in the time dimension; The data fusion unit integrates the cleaned, standardized and synchronized multi-source data to generate a unified data structure; The output formatting unit converts the fused data into a predefined standard format to facilitate processing and analysis by subsequent modules.

4. A portable respiratory function rehabilitation training device according to claim 1, characterized in that: The intelligent analysis module includes the following components: Feature extraction unit, extracts key features from the fused data to provide high-quality input for subsequent analysis; The deep learning model unit uses a pre-trained neural network model to perform in-depth analysis on the extracted features and identify complex data patterns; The time series analysis unit uses time series analysis techniques to capture the time dependency and changing trends of data; The breathing pattern recognition unit combines deep learning and timing analysis results to identify and classify different breathing patterns; A health status assessment unit that assesses the user's current health status based on the identified breathing pattern and other physiological indicators; The rehabilitation trend prediction unit integrates the user's historical health data and current assessment results to predict the user's rehabilitation trend and potential risks.

5. The portable respiratory function rehabilitation training device according to claim 1, characterized in that: The training scheme generation module includes the following components: User profile management unit, which stores and updates the user's long-term health data, training history and personal preferences, and provides background information for program generation; The training goal setting unit sets and manages short-term and long-term training goals according to medical advice and user wishes; The real-time status input unit receives and processes the real-time user status data from the intelligent analysis module to ensure that the solution matches the current situation; The training program generator takes into account the user profile, training goals and real-time status to generate preliminary personalized training programs; The program adaptive adjustment unit dynamically adjusts and optimizes the training program based on the user's real-time feedback and training progress to ensure its continued effectiveness; The training instruction output unit converts the generated training plan into specific and executable training instructions and passes them to the real-time feedback and interaction module.

6. A portable respiratory function rehabilitation training device according to claim 1, characterized in that: The real-time feedback and interaction module includes the following components: A training instruction receiving unit receives specific instructions from the training program generating module and provides basic content for feedback generation; Multimodal feedback generator, which generates feedback content suitable for different sensory channels based on training instructions and user real-time status; Visual feedback unit, which displays visual information including training guidance, progress bar, and performance statistics through a graphical interface; Auditory feedback unit, which uses voice prompts and sound effects to provide real-time training guidance and encouraging feedback; The tactile feedback unit provides training rhythm guidance and completion prompts through tactile signals such as vibration; The user input processing unit receives and analyzes user operations and voice commands to achieve two-way interaction.

7. The portable respiratory function rehabilitation training device according to claim 1, characterized in that: The performance evaluation and optimization module includes the following components: Long-term data recording unit, which continuously collects and stores the user's training data, physiological indicators and health status, providing comprehensive historical data support for subsequent analysis; The effect evaluation engine comprehensively analyzes the user's rehabilitation progress and training effect based on the recorded long-term data and generates a quantitative evaluation report; The health trend analysis unit uses time series analysis technology to conduct in-depth analysis of long-term changes in users' health status and identify key health trends and patterns; Reinforcement learning optimizer, which models the training process as a reinforcement learning problem and continuously optimizes the training strategy based on evaluation results and health trends; The strategy update unit receives the optimized strategy and converts it into specific training program adjustment suggestions to achieve dynamic update of the training strategy; The personalized parameter adjustment unit fine-tunes the optimized training strategy according to the user's individual characteristics and feedback to ensure its adaptability and personalization.

8. A portable respiratory function rehabilitation training method, implemented based on a portable respiratory function rehabilitation training device according to any one of claims 1 to 7, characterized in that: The following steps are involved: Collect user's physiological parameters, posture information and environmental parameters in real time; Clean, standardize and time-synchronize the collected data and merge them into a unified format; Extract key features of the user's respiratory function from the fused data, identify breathing patterns, and conduct real-time health status assessment; Generate a preliminary personalized training plan based on the user's personal profile, training goals and real-time health status; Provide users with instant training guidance and feedback on current progress through multimodal feedback; Continuously record users' long-term training data and generate quantitative training effect evaluation reports; Based on the evaluation results, the training strategy is updated and fine-tuned to achieve adaptive optimization and personalized iteration of the training plan.

Images