Intelligent equipment for monitoring cardio-pulmonary function of heart failure patient

Abstract

The invention relates to the technical field of medical instruments, and discloses an intelligent device for monitoring the cardio-pulmonary function of a heart failure patient, the intelligent device comprises a multi-mode sensor module, an AI analysis module, a biological feedback module and a wearable carrier, the multi-mode sensor module is used for collecting heart rate, respiratory rate, oxyhemoglobin saturation and body movement data in real time, and the AI analysis module is used for analyzing the heart rate, the respiratory rate, the oxyhemoglobin saturation and the body movement data; monitoring modes can be dynamically switched in rest, motion and sleep scenes; the AI analysis module is used for detecting abnormity based on real-time data and predicting a heart failure deterioration risk in combination with historical data; and the biological feedback module is used for providing personalized intervention in a vibration or audio mode according to the analysis result. The heart rate, the respiration rate, the blood oxygen saturation and the body movement data are collected at the same time through the multi-mode sensor module, sleep apnea or periodic respiration is recognized in combination with the AI analysis module, the defect of separation of cardiopulmonary function monitoring and sleep mode analysis in the prior art is overcome, and therefore early intervention and treatment adjustment are supported.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and particularly to an intelligent device for monitoring the cardiopulmonary function of heart failure patients. Background Art

[0002] Heart failure is a chronic progressive disease, often accompanied by abnormal cardiopulmonary function, and continuous monitoring is required to evaluate the condition and guide treatment. In the prior art, a variety of devices for monitoring the cardiopulmonary function of heart failure patients have been developed. For example, implantable devices such as CardioMEMS monitor pulmonary artery pressure through sensors implanted in the pulmonary artery and wirelessly transmit the data to an external system for detecting the risk of fluid overload; wearable devices such as the Sensinel system use non-invasive sensors to collect multiple parameters such as heart rate, respiratory rate, and chest impedance, which are suitable for daily monitoring in a home environment. In addition, some studies have explored cardiopulmonary exercise testing (CPET) to measure the maximum oxygen uptake (VO2max) to evaluate the exercise endurance of heart failure patients, and some other devices combine artificial intelligence (AI) technology to analyze electrocardiogram (ECG) data to predict the deterioration trend of heart failure. These technologies have, to varying degrees, met the needs of monitoring the cardiopulmonary function of heart failure patients.

[0003] However, the prior art has several limitations. Implantable devices such as CardioMEMS need to be implanted through invasive surgery, with limited scope of application, and only focus on a single parameter (pulmonary artery pressure), unable to comprehensively reflect the cardiopulmonary function status. Wearable devices such as the Sensinel system can monitor multiple parameters, but lack targeted analysis of heart failure-related complications such as sleep disordered breathing (SDB), and the incidence of SDB in heart failure patients is as high as 50 - 80%. In addition, the functions of existing devices are mostly limited to data collection and alarms, and fail to provide real-time intervention measures when detecting abnormalities, resulting in patients being difficult to improve their conditions through self-adjustment. Although the application of AI technology has improved the prediction ability, it mostly focuses on short-term risk assessment and lacks personalized threshold setting based on long-term trends, limiting its utility in the management of chronic heart failure. These deficiencies indicate that the prior art still needs to be improved in terms of monitoring comprehensiveness, intervention ability, and personalized management. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides an intelligent device for monitoring the cardiopulmonary function of heart failure patients, which solves the problem that existing wearable devices lack targeted analysis of heart failure-related complications such as sleep disordered breathing (SDB).

[0005] To achieve the above object, the present invention is realized through the following technical solutions: An intelligent device for monitoring the cardiopulmonary function of heart failure patients, including a multi-mode sensor module, an AI analysis module, a biofeedback module, and a wearable carrier, wherein: The multi-mode sensor module is used to collect real-time heart rate, respiratory rate, blood oxygen saturation, and body movement data, and supports dynamic switching of monitoring modes in rest, exercise, and sleep scenarios; The AI analysis module, connected to the multi-mode sensor module, is used to detect anomalies based on real-time data and predict the risk of heart failure deterioration in combination with historical data; The biofeedback module, connected to the AI analysis module, is used to provide personalized interventions in the form of vibration or audio according to the analysis results to guide the patient to adjust breathing or activities; The wearable carrier is used to integrate the above modules and is suitable for long-term wearing by patients.

[0006] Preferably, the multi-mode sensor module includes a photoplethysmography (PPG) sensor, an accelerometer, and an impedance sensor, where: The PPG sensor is used to collect heart rate and blood oxygen saturation data; The accelerometer is used to collect body movement data to identify activity states and sleep patterns; The impedance sensor is used to collect chest impedance data to monitor respiratory rate and changes in fluid overload.

[0007] Preferably, the multi-mode sensor module can identify sleep apnea or periodic breathing in the sleep scenario and transmit the identification results to the AI analysis module for analysis.

[0008] Preferably, the AI analysis module includes an embedded microprocessor and an algorithm based on a long short-term memory network, where: The algorithm is used to analyze historical data over several weeks to months, generate personalized health thresholds, and predict the time of heart failure deterioration; The AI analysis module can dynamically adjust the threshold for anomaly detection according to the patient's activity state.

[0009] Preferably, the biofeedback module includes a vibration motor and a Bluetooth audio output unit, where: The vibration motor is used to prompt the patient to adjust breathing when shortness of breath or a drop in blood oxygen is detected; The Bluetooth audio output unit is used to guide the patient to adjust the sleeping position or breathing pattern through audio in the sleep scenario.

[0010] Preferably, it further includes a data storage and communication module, which includes a flash memory chip and a low-power Bluetooth unit, and is used to store historical data and transmit the monitoring results to external devices.

[0011] Preferably, the wearable carrier is a chest patch or smart clothing, made of flexible circuits and breathable materials to ensure wearing comfort and long-term use stability.

[0012] Preferably, the AI ​​analysis module can predict the risk of fluid overload by analyzing heart rate variability HRV and respiratory rate trends, and trigger the biofeedback module to intervene when the predicted result exceeds a personalized threshold.

[0013] Preferably, it also includes a self-powered unit, which powers the device through body temperature or motion kinetic energy to reduce the need for external charging.

[0014] Preferably, it also supports multi-user data management, and data can be stored and shared remotely by connecting to a mobile phone application, which is suitable for home or community use scenarios.

[0015] The present invention provides an intelligent device for monitoring the cardiopulmonary function of patients with heart failure. It has the following beneficial effects: 1. The present invention uses a multi-mode sensor module to simultaneously collect heart rate, respiratory rate, blood oxygen saturation and body movement data, and combines with an AI analysis module to identify sleep apnea or periodic breathing, filling the deficiency of the existing technology that cardiopulmonary function monitoring and sleep pattern analysis are separated; in the sleep scene, the PPG sensor and accelerometer work together to detect the time characteristics of blood oxygen drop and chest movement pause, and generate an hourly pause index (AHI), providing doctors with a more comprehensive basis for disease assessment, thereby supporting early intervention and treatment adjustment.

[0016] 2. The present invention is equipped with a biofeedback module, which can guide the patient to adjust his breathing or activity status in real time through vibration or audio according to the abnormalities detected by the AI ​​analysis module (such as respiratory rate exceeding 25 times / minute or blood oxygen below 90%); trigger personalized intervention through a vibration motor or a Bluetooth audio output unit, such as prompting to turn over in sleep apnea, or suggesting to slow down in exercise overload; this active intervention mechanism utilizes real-time changes in heart rate variability (HRV) or breathing patterns, and aims to reduce the risk of hypoxic events or fluid overload through patient behavior adjustments, thereby improving the effectiveness of daily management.

[0017] 3. The present invention uses an AI analysis module to analyze historical data from several weeks to several months to generate personalized health thresholds and predict the time of heart failure deterioration, which exceeds the limitations of short-term abnormality detection in existing technologies (such as the AI ​​prediction of VA research is only one week in advance); the device uses a long short-term memory network (LSTM) algorithm to process the time series of heart rate, respiratory rate and impedance. If the baseline respiratory rate rises by 5 times / minute and the impedance decreases by 10%, the risk of deterioration can be warned 7 days in advance; the personalized threshold is adjusted according to the patient's baseline (such as age, heart function grade) to avoid the false positives or omissions of the general threshold, providing doctors with a more accurate long-term management basis and optimizing the timing of drug or lifestyle interventions.

[0018] 4. The present invention supports dynamic monitoring in rest, exercise, and sleep scenarios through a multi-mode sensor module, and adjusts the data acquisition mode and analysis parameters according to the activity status, solving the limitation of single-scenario monitoring in the prior art; the accelerometer identifies the patient's status (such as a walking frequency of 2 - 3 Hz), and the AI module adjusts the threshold accordingly (such as a heart rate upper limit of 130 beats per minute during exercise and 100 beats per minute during rest), ensuring that the monitoring results match the actual load, thereby improving the applicability and reliability of the data in different life scenarios.

[0019] 5. The present invention uses a self-powered unit to generate energy using body temperature or kinetic energy of movement (such as 6 mW of thermoelectric power generation and 1.5 mW of piezoelectric power generation), contributing approximately 150 mWh per day. In combination with a 500 mAh battery, the battery life is extended from 48 hours to 52 hours, reducing the frequency of external charging. The energy harvesting design of the thermoelectric module and piezoelectric sheet reduces the patient's dependence on device maintenance, especially suitable for elderly heart failure patients with limited activity, and extends the feasibility of continuous monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is the overall system architecture diagram of an intelligent device for monitoring the cardiopulmonary function of heart failure patients according to the present invention; Figure 2 is the multi-mode sensor module architecture diagram of an intelligent device for monitoring the cardiopulmonary function of heart failure patients according to the present invention; Figure 3 is the AI analysis module architecture diagram of an intelligent device for monitoring the cardiopulmonary function of heart failure patients according to the present invention; Figure 4 is the biofeedback module architecture diagram of an intelligent device for monitoring the cardiopulmonary function of heart failure patients according to the present invention; Figure 5 is the data storage and communication module architecture diagram of an intelligent device for monitoring the cardiopulmonary function of heart failure patients according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0022] Please refer to the attached Figure 1 - attached Figure 5 , an embodiment of the present invention provides an intelligent device for monitoring the cardiopulmonary function of heart failure patients, including a multi-mode sensor module, an AI analysis module, a biofeedback module, and a wearable carrier, wherein: A multi-mode sensor module for real-time collection of heart rate, respiratory rate, blood oxygen saturation, and body movement data, and supporting dynamic switching of monitoring modes in rest, exercise, and sleep scenarios; An AI analysis module, connected to the multi-mode sensor module, for detecting abnormalities based on real-time data and predicting the risk of heart failure deterioration in combination with historical data; A biofeedback module, connected to the AI analysis module, for providing personalized interventions in the form of vibration or audio according to the analysis results to guide the patient to adjust breathing or activities; A wearable carrier for integrating the above modules and suitable for long-term wearing by patients.

[0023] Specifically, the intelligent device realizes multi-parameter synchronous collection through the multi-mode sensor module during actual use. For example, when the patient is sleeping at night, the device collects heart rate and blood oxygen saturation data at a frequency of 2 times per second, and simultaneously records body movement and chest impedance changes at a frequency of 1 time per minute. When the patient is walking during the day, the device automatically recognizes the exercise state and switches to the high-frequency collection mode (5 times per second), focusing on monitoring the heart rate and respiratory rate to capture the cardiopulmonary response under exercise load. The AI analysis module is built into the microprocessor of the device and runs the LSTM model using the embedded TensorFlowLite framework. The model is pre-trained in the cloud with 100,000 sets of heart failure patient data and has an abnormal detection accuracy of 95%. During real-time analysis, the AI module processes data every 5 minutes. When the heart rate exceeds 120 beats per minute and the blood oxygen is lower than 90%, it is judged as abnormal. Subsequently, the biofeedback module is activated. For example, it emits 3 short vibrations at 2-second intervals through a vibration motor to prompt the patient to slow down activities, or plays the pre-recorded voice instruction of "Please take 5 deep breaths" through a Bluetooth headset. The wearable carrier is made of medical-grade silicone, with dimensions of 8 cm × 5 cm × 0.5 cm and a weight of about 30 g. It is attached to the lower part of the patient's sternum and is built-in with a flexible battery to support continuous operation for 48 hours. The device is also equipped with LED indicators, green for normal and red flashing for abnormal, facilitating the patient to intuitively understand the status.

[0024] Please refer to the appendix Figure 2 . The multi-mode sensor module includes a photoplethysmography (PPG) sensor, an accelerometer, and an impedance sensor, where: The PPG sensor is used to collect heart rate and blood oxygen saturation data; The accelerometer is used to collect body movement data to identify activity states and sleep patterns; The impedance sensor is used to collect chest impedance data to monitor respiratory rate and changes in fluid overload.

[0025] Specifically, the implementation of the multi-mode sensor module includes the following details: The PPG sensor uses the MAX30102 model, which integrates red and infrared LEDs with wavelengths of 660nm and 880nm respectively. It collects pulse wave signals through reflective measurement, calculates heart rate (range 30-220 times / minute, error ±2 times / minute) and blood oxygen saturation (range 70-100%, error ±2%). The accelerometer uses the MPU-6050 three-axis sensor with a sensitivity of ±2g, which can detect changes in patient position (such as standing, lying) and small movements during sleep (such as chest rise and fall). The sampling frequency is adjustable (1-100Hz) and is set to 10Hz in sleep mode to save power. The impedance sensor is based on the AD5933 chip, which measures the change in chest resistance (range 10-100Ω) by applying a 10kHz AC signal, and calculates the respiratory rate (range 5-40 times / minute, error ±1 time / minute) in combination with the respiratory cycle, and estimates the degree of pleural fluid accumulation through impedance increment (an increment exceeding 20% ​​is considered abnormal). These sensors are connected to the main MCU (model STM32F4) through the I2C interface, and the data is processed by a low-pass filter and stored in the buffer to ensure signal stability and anti-interference ability.

[0026] Please see attached Figure 2 ,The multi-mode sensor module can identify sleep apnea or periodic breathing in the sleep scenario, and transmit the recognition results to the AI ​​analysis module for analysis.

[0027] Specifically, in the sleep scenario, the multi-mode sensor module uses the accelerometer and PPG sensor to work together to identify sleep apnea or periodic breathing. For example, when the accelerometer detects that the chest movement stops for more than 10 seconds, and the PPG sensor records that the blood oxygen saturation drops by more than 4% (such as from 95% to 90%), the device determines it as a sleep apnea event. If such an event occurs more than 5 times within 1 hour, the AI ​​module marks it as "moderate sleep disordered breathing" and records the timestamp and duration (accurate to seconds). The identification of periodic breathing (such as Cheyne-Stokes breathing) is based on the periodic changes in respiratory rate fluctuations. The AI ​​module analyzes the respiratory signal through fast Fourier transform (FFT). If a significant frequency peak of 0.01-0.03Hz is detected, it is confirmed as periodic breathing. The recognition result is transmitted to the AI ​​module in JSON format (for example, {"event":"apnea","duration":"12s","timestamp":"03:15:27"}) for subsequent risk assessment. After waking up, the patient can view the sleep report through the App, which includes the total number of pauses, average blood oxygen drop and suggestions (such as "it is recommended to consult a doctor").

[0028] Please see attached Figure 3, the AI analysis module includes an embedded microprocessor and an algorithm based on long short-term memory network, where: The algorithm is used to analyze historical data from several weeks to several months, generate personalized health thresholds, and predict the time of heart failure deterioration; The AI analysis module can dynamically adjust the threshold of anomaly detection according to the patient's activity status.

[0029] Specifically, the core of the AI analysis module is an embedded system based on the STM32F4 microprocessor, with an operating clock frequency of 168 MHz, equipped with 256 KB RAM and 1 MB Flash storage space, supporting real-time data processing and model inference. The LSTM algorithm consists of three layers of neural networks, with 50 neurons in each layer. The input layer receives time series data from multi-modal sensors (each group of data includes heart rate, respiratory rate, etc. within a 60-second window), and the output layer generates an anomaly probability (in the range of 0 - 1, with a threshold of 0.8 considered as an anomaly). Model training is completed in the cloud, using the Keras library in Python. The dataset includes labeled data of 500 heart failure patients (NYHA classification I - IV). After training, the model is compressed to 50 KB in size and adapted for embedded operation. The long-term prediction function is achieved by analyzing 4-week historical data. For example, if the respiratory rate baseline rises from 15 breaths per minute to 20 breaths per minute, and the average blood oxygen level drops by 2%, the model predicts an 85% deterioration probability after 7 days and generates a trend graph (the horizontal axis is days, and the vertical axis is the parameter value). The dynamic threshold adjustment is achieved according to the activity status. For example, the heart rate threshold is 100 beats per minute during rest and rises to 130 beats per minute during exercise, ensuring a false alarm rate of less than 5%.

[0030] Please refer to the appendix Figure 4 , the biofeedback module includes a vibration motor and a Bluetooth audio output unit, where: The vibration motor is used to prompt the patient to adjust breathing when shortness of breath or a drop in blood oxygen is detected; The Bluetooth audio output unit is used to guide the patient to adjust the sleeping position or breathing pattern through audio in the sleep scenario.

[0031] Specifically, the specific design of the biofeedback module includes: the vibration motor selects an eccentric rotor motor with a diameter of 10 mm, a working voltage of 3 V, an amplitude of 0.5 g, and a programmable vibration mode (such as a short vibration for 0.5 seconds and a long vibration for 2 seconds). For example, when the AI detects that the respiratory rate exceeds 25 times per minute and lasts for 30 seconds, the vibration motor runs in the sequence of "short vibration - stop - short vibration - stop - long vibration" to prompt the patient to perform abdominal breathing (inhale for 4 seconds and exhale for 6 seconds) until the respiratory rate drops below 20 times per minute. The Bluetooth audio output unit is based on the nRF52832 chip, supports the BLE5.0 protocol, and can play pre-recorded MP3 voice files (stored in a 16MB flash memory) after pairing with the headphones, such as "Please take a slow deep breath" or "Adjust to the left lateral position". In the sleep scenario, if a respiratory pause exceeding 15 seconds is detected, the headphones play a low-volume prompt tone (50 decibels) for 5 seconds to avoid waking up the patient while promoting self-adjustment. The power consumption of this module is controlled below 5 mW, and the response time is less than 1 second to ensure real-time performance.

[0032] Please refer to the appendix Figure 5 , and also includes a data storage and communication module. The data storage and communication module includes a flash chip and a low-power Bluetooth unit, which are used to store historical data and transmit the monitoring results to external devices.

[0033] Specifically, the specific implementation of the data storage and communication module includes: the flash chip uses the W25Q128 model with a capacity of 128 Mbit, which can store 30 days of monitoring data (about 1 KB is recorded per minute). The data is stored indexed by timestamp, such as "2025-03-14 08:30:00, HR:85, RR:16, SpO2:95", and offline query is supported. The low-power Bluetooth unit is based on the nRF52832 chip, with a transmission rate of 250 kbps and a coverage range of 10 meters. The data is synchronized with the mobile App every 5 minutes (about 50 KB each time). During the synchronization process, the device uses AES-128 encryption to protect data privacy, and the App generates a visual report, including a heart rate trend curve (the horizontal axis is hours and the vertical axis is times per minute), a blood oxygen fluctuation graph, and a list of abnormal events (such as "03:15, apnea, lasting 12 seconds"). If the network is available, the data is further uploaded to the cloud server (based on AWS IoT), and the doctor can remotely access it through the web interface to adjust the thresholds (such as changing the blood oxygen alarm from 90% to 92%), and the adjustment instructions are sent to the device in real time.

[0034] Please refer to the appendix Figure 1 , and the wearable carrier is a chest patch or smart clothing, which is made of flexible circuits and breathable materials to ensure wearing comfort and long-term use stability.

[0035] Specifically, the design details of the wearable carrier are as follows: The chest patch is made of medical silicone substrate, with dimensions of 8 cm × 5 cm × 0.5 cm, weighing 30 g. The surface is coated with a waterproof coating (IPX5 level), resistant to sweat and moisture. The internal flexible circuit board is based on polyimide (PI) material, with a thickness of 0.1 mm, supporting a bending radius of 5 mm to ensure a good fit to the chest curve. The smart clothing is in the form of a tight-fitting top, with sensors embedded in the chest area. Conductive fibers (resistance < 1 Ω / cm) are used to connect the modules. The clothing fabric is a cotton-polyester blend (ratio 60:40), with a breathability of 200 g / m² / 24 h. The patch is fixed with medical double-sided tape (adhesive lasts for 7 days), and the clothing is designed with a zipper for easy putting on and taking off. The device is equipped with a 3.7V, 500 mAh flexible lithium battery, which can support 48 hours of operation when fully charged, and the charging time is 2 hours (through the USB-C interface). When the patient wears it, the patch is placed 2 cm below the sternum, or the clothing is aligned with the middle of the pectoralis major muscle to ensure good contact between the sensor and the skin, with a signal attenuation of less than 5%.

[0036] Please refer to the appendix Figure 3 , the AI analysis module can predict the risk of fluid overload by analyzing the heart rate variability HRV and the respiratory rate trend, and trigger the biofeedback module to intervene when the prediction result exceeds the personalized threshold.

[0037] Specifically, the specific analysis process of the AI analysis module is as follows: The heart rate variability (HRV) is calculated from the pulse intervals (R-R intervals) collected by the PPG sensor, and the time domain method (SDNN index, unit ms) is used to evaluate the autonomic nerve function. For example, an SDNN below 20 ms indicates a worsening of heart failure. The respiratory rate trend is calculated from the impedance sensor data, and the daily average value is recorded as a line graph (the horizontal axis is days, and the vertical axis is times / minute). The risk of fluid overload is based on the combined analysis of chest impedance and blood oxygen data. For example, a 15% decrease in impedance (such as from 50 Ω to 42.5 Ω) and an average blood oxygen level below 93% trigger a risk assessment. The AI module runs a trend analysis every 24 hours. If the HRV drops by 10% and the respiratory rate increases by 5 times / minute for 3 consecutive days, the predicted probability of deterioration rises to 80%, and an alarm "The risk will increase in 5 days" is generated. The biofeedback module is then activated. For example, the vibration motor emits 5 short vibrations at 1-second intervals to prompt the patient to rest and contact the doctor, and at the same time, the App pushes a detailed report (including the HRV curve and the impedance change table).

[0038] Please refer to the appendix Figure 1 , and also includes a self-power supply unit, which powers the device through body temperature or kinetic energy of movement to reduce the need for external charging.

[0039] Specifically, for thermoelectric power generation: a TEG1-1263-4.3 thermoelectric power generation module (size 15mm×15mm, weight 5g) is used and placed on the side of the patch close to the skin to generate electricity by utilizing the temperature difference between the chest skin and the environment (average 3 - 5°C). At a typical temperature difference of 4°C, the output voltage is 0.4V and the power is about 6mW (based on Bi2Te3 material, efficiency 6%), which is converted to 3.3V through a boost circuit (TPS61200, efficiency 90%). The module operates for 24 hours a day, generating about 144mWh of energy (6mW×24h), which can support the device to operate in the standby mode (power consumption 2mW) for 72 hours, or in the low-frequency monitoring mode (sampling rate once per minute, power consumption 5mW) for 28 hours. To improve the efficiency, a heat insulation layer (polyurethane foam, thickness 2mm) is designed on the outer layer of the patch to enhance the temperature difference stability.

[0040] Piezoelectric power generation: A PZT-5H piezoelectric ceramic sheet (size 10mm×10mm, thickness 0.5mm) is used and embedded in the sternum area of the patch to generate electricity by utilizing the micro-vibration of the chest during walking or breathing. During walking (frequency 2Hz, amplitude 0.05g), the output power is about 15μW / cm² and the total power is 1.5mW (area 10cm²); during breathing (frequency 0.3Hz, amplitude 0.02g), the output is about 5μW / cm² and the total power is 0.5mW. The patient generates 2.7mWh of energy (1.5mW×0.5h) by walking for 30 minutes a day and 7.2mWh (0.5mW×24h) by breathing for 24 hours, with a total of about 10mWh, which is stored in a 20μF supercapacitor and can support Bluetooth transmission (power consumption 5mW) to operate for about 2 minutes.

[0041] Power management: The BQ25504 energy management chip is adopted, which integrates the MPPT (maximum power point tracking) function. The thermoelectric and piezoelectric energies are rectified respectively and stored in a 500mAh lithium battery (3.7V, total capacity 1850mWh). The self-powered supply contributes about 154mWh (144 + 10) per day, accounting for 8.3% of the total energy consumption, and can extend the battery life from 48 hours to 52 hours. If the patient's activity level increases (walking for 2 hours), the piezoelectric contribution rises to 15mWh and the total battery life can reach 54 hours. The App displays the power supply status (such as "Thermoelectric has supplied 120mWh and the battery remains 80%"), and the patient can choose the charging time according to the prompt.

[0042] Please refer to the attachment Figure 1 and also includes support for multi-user data management, which stores and remotely shares data in separate accounts by connecting to a mobile application, and is applicable to home or community usage scenarios.

[0043] Specifically, the implementation details of multi-user data management are as follows: the device is paired with the mobile phone App via Bluetooth. The App supports up to 5 user accounts, each account is bound to a unique ID (such as "User1: Mr. Zhang"), and the data is stored in the flash partition (25MB is allocated for each user) by account. When patient A wears it, the App records his heart rate and blood oxygen data. When switching to patient B, manually select "Account B", and the device automatically loads B's personalized threshold (such as heart rate upper limit 110 times / minute). When data is synchronized, the App generates independent reports, such as "Mr. Zhang: 03-14, average heart rate 85, 2 abnormal events"; "Ms. Li: 03-14, average blood oxygen 94%, sleep pause 3 times". The remote sharing function is realized through the cloud. After the patient authorizes, family members can view real-time data (refresh interval 5 minutes), and doctors can adjust parameters through the Web end (such as changing the respiratory rate threshold from 20 to 22 times / minute), and the adjustment instructions are transmitted to the device via 256-bit encryption. This function is particularly suitable for nursing home scenarios, and administrators can manage the monitoring status of multiple patients in batches.

[0044] The following is an introduction in conjunction with specific embodiments: Example 1: Sleep monitoring and biofeedback optimization This embodiment focuses on cardiopulmonary function monitoring and real-time intervention during sleep in patients with heart failure, and is particularly suitable for patients with sleep-disordered breathing (SDB).

[0045] Hardware: The device is in the form of a chest patch (size 7cm×4cm×0.5cm, weight 25g), which includes a PPG sensor (MAX30102, collecting heart rate and blood oxygen), an accelerometer (MPU-6050, detecting body movement and breathing fluctuations), a vibration motor (diameter 8mm, amplitude 0.4g) and a Bluetooth headset module (nRF52832).

[0046] In sleep mode, the PPG sensor monitors blood oxygen and heart rate at a frequency of 2Hz, and the accelerometer detects chest movement at 10Hz.

[0047] The AI ​​module (STM32F4, running the LSTM algorithm) analyzes the data. When the blood oxygen drops by 4% (such as 95% to 91%) and the chest movement stops for 10 seconds, it is judged as sleep apnea, and the event is recorded and the apnea index (AHI) per hour is calculated.

[0048] The biofeedback module triggered the intervention: slight vibration (0.5 s × 3 times) prompted the patient to turn over, and if the pause lasted for 15 s, the earphones played a low-volume reminder tone (“Please take a deep breath”, 40 decibels).

[0049] Data processing: A sleep status report is generated every 5 minutes and stored in a 16MB flash memory. The AHI curve (horizontal axis is hours, vertical axis is the number of pauses) and the lowest blood oxygen value are displayed through the App the next day.

[0050] Power supply: 3.7V, 400mAh battery, supports 36-hour operation, and thermoelectric power generation (TEG1-1263-4.3, output 5mW) extends the operation to 40 hours.

[0051] Application scenario: Patient Zhang, 65 years old, with heart failure NYHA II level, wears the device to sleep for 8 hours every night. One night, the AHI reached 10 times per hour, the lowest blood oxygen level was 88%, the device vibrated 3 times and then the patient turned over, and the blood oxygen level recovered to 92%. The next day, the doctor adjusted the CPAP treatment according to the report.

[0052] Example 2: Dynamic monitoring type for exercise scenarios This example is for monitoring the cardiopulmonary function of heart failure patients during daily activities (such as walking), emphasizing dynamic scenario adaptation and personalized threshold adjustment.

[0053] Hardware: The device is in the form of smart clothing (a tight-fitting upper garment with sensors embedded in the chest), including a PPG sensor (collecting heart rate and blood oxygen), an impedance sensor (AD5933, monitoring respiratory rate and thoracic fluid), an accelerometer (detecting the motion state), and a vibration motor.

[0054] The accelerometer identifies the walking state (frequency 1 - 3Hz, amplitude 0.1g), the device switches to the exercise mode, the sampling rate increases to 5Hz, and it preferentially monitors the heart rate (range 50 - 150 beats per minute) and the respiratory rate (10 - 40 breaths per minute).

[0055] The AI module dynamically adjusts the threshold according to the patient's baseline (preset heart rate upper limit of 120 beats per minute). If the heart rate reaches 125 beats per minute and the respiratory rate exceeds 30 breaths per minute, it is judged as overloading.

[0056] The biofeedback module starts vibrating (1 second × 5 times) to prompt the patient to slow down or rest, and the App synchronously displays "Please pause walking".

[0057] Data processing: Real-time records the exercise duration and cardiopulmonary load, generates a daily report (heart rate peak value, respiratory rate trend), and transmits it to the mobile phone via Bluetooth.

[0058] Power supply: 500mAh battery supports 48 hours, and piezoelectric power generation (PZT-5H, 1.5mW during walking) supplements energy, extending the battery life by 15 minutes per hour of walking.

[0059] Application scenario: Patient Li, 70 years old, with heart failure NYHA III level, walks for 20 minutes every day. One day, the heart rate rose to 130 beats per minute and the patient was short of breath. After the device vibrated to prompt, the patient rested for 5 minutes, and the heart rate dropped to 100 beats per minute, avoiding deterioration.

[0060] Example 3: Long-term trend prediction and home sharing type This embodiment focuses on long-term heart failure risk prediction and multi-user home management, and is suitable for multiple heart failure patients in the community or at home.

[0061] Hardware: The device is a chest patch (8cm×5cm×0.5cm) that integrates a PPG sensor, an impedance sensor, an accelerometer, 128MB flash memory, and a Bluetooth module (nRF52832).

[0062] The data (heart rate, respiratory rate, blood oxygen) were continuously monitored for 4 weeks, and the AI ​​module (LSTM algorithm, 50KB model) analyzed the trend. If the baseline respiratory rate increased by 5 times / minute (15 to 20 times / minute) and the impedance decreased by 10% (50Ω to 45Ω), the probability of deterioration after 7 days was predicted to be 80%.

[0063] Support multi-user switching: The App is bound to 3 accounts (such as "Father Zhang" and "Mother Zhang"), and the data is stored separately, with a capacity of 25MB for each user.

[0064] Cloud sharing: Doctors view trend charts through the web (the horizontal axis is days, the vertical axis is parameter values), and family members receive real-time alerts (such as "Mr. Zhang's breathing rate is abnormal").

[0065] Data processing: Generate trend reports every week (PDF format, including HRV and fluid overload index), and doctors can remotely adjust thresholds (such as raising the lower limit of blood oxygen from 90% to 92%).

[0066] Power supply: 500mAh battery supports 48 hours, thermoelectric power generation (6mW) extends to 52 hours.

[0067] Application scenario: Mr. and Mrs. Zhang are both patients with heart failure. They wear the same device and take turns using it. In a certain month, Mr. Zhang's respiratory rate trend increases. The device predicts the risk, and the doctor prescribes diuretics in advance, reducing the risk of hospitalization.

[0068] Example 4: Self-powered and low-power optimized This embodiment focuses on self-powered design and low-power operation, and is suitable for elderly patients with low activity and high dependence.

[0069] Hardware: Patch device (6cm×4cm×0.5cm, 20g) containing a PPG sensor (1mW in low-power mode), an accelerometer, a thermoelectric module (TEG1-1263-4.3), a piezoelectric sheet (PZT-5H), and a supercapacitor (20μF).

[0070] Thermoelectric power generation uses the temperature difference in the chest (3-5°C) to generate 6mW of power, and piezoelectric power generation uses breathing (0.5mW) or walking (1.5mW) to supplement energy, totaling 150mWh per day.

[0071] Low-power mode: The sensor sampling rate is reduced to once per minute, the AI analysis runs once per hour, and the power consumption is controlled at 3 mW.

[0072] Biofeedback is triggered only in case of abnormalities (such as blood oxygen level below 90%, vibration for 0.5 seconds × 3 times).

[0073] Data processing: Data is stored in an 8 MB flash memory and synchronized to the App daily, displaying the power supply status ("Thermoelectric contribution 60%") and cardiopulmonary parameters.

[0074] Power supply: A 400 mAh battery supports 36 hours, self-power supply is extended to 48 hours, and walking for 1 hour adds an additional 4 hours.

[0075] Application scenario: Patient Wang, 80 years old, mainly bedridden. The device operates by thermoelectric and respiratory power generation. One day, the blood oxygen level dropped to 88%. After the vibration prompt, the family members assisted in adjusting the body position, and there was no problem with the battery life.

[0076] Example 5: Comprehensive monitoring and personalized intervention type This example integrates all functions to provide comprehensive monitoring and personalized management, suitable for heart failure patients with complex conditions.

[0077] Hardware: Smart clothing (modular design on the chest), including PPG, impedance, accelerometer, vibration motor, headphone module, 128 MB flash memory, and thermoelectric + piezoelectric unit.

[0078] Full-scenario monitoring: Detect AHI during sleep, monitor heart rate peak during exercise, and evaluate HRV during rest.

[0079] AI prediction: Analyze 4-week data to generate personalized thresholds (such as heart rate upper limit of 115 beats per minute), predict deterioration, and push alerts.

[0080] Biofeedback: The headphone guides turning over during sleep, the vibration prompts rest during exercise, and the App gives voice reminders in case of abnormalities.

[0081] Data processing: The App generates comprehensive reports (sleep quality, cardiopulmonary load, risk trend), and doctors can remotely adjust parameters.

[0082] Power supply: A 500 mAh battery supports 48 hours, and self-power supply is extended to 54 hours.

[0083] Application scenario: Patient Zhao, 60 years old, with heart failure NYHA IV. Worn all day long. One day, excessive exercise was detected and a rest prompt was given. The sleep apnea was corrected, and the doctor adjusted the medication according to the trend.

[0084] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will appreciate that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent device for monitoring the cardiopulmonary function of patients with heart failure, characterized in that: It includes a multi-mode sensor module, an AI analysis module, a biofeedback module and a wearable carrier, among which: The multi-mode sensor module is used to collect heart rate, respiratory rate, blood oxygen saturation and body movement data in real time, and supports dynamic switching of monitoring modes in rest, exercise and sleep scenarios; The AI ​​analysis module is connected to the multi-mode sensor module and is used to detect abnormalities based on real-time data and predict the risk of worsening heart failure in combination with historical data; The biofeedback module is connected to the AI ​​analysis module and is used to provide personalized intervention through vibration or audio according to the analysis results to guide the patient to adjust breathing or activities; The wearable carrier is used to integrate the above modules and is suitable for patients to wear for a long time.

2. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: The multi-mode sensor module includes a photoplethysmography (PPG) sensor, an accelerometer, and an impedance sensor, wherein: The PPG sensor is used to collect heart rate and blood oxygen saturation data; The accelerometer is used to collect body motion data to identify activity status and sleep patterns; The impedance sensor is used to collect chest impedance data to monitor respiratory rate and fluid overload changes.

3. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: The multi-mode sensor module can identify sleep apnea or periodic breathing in a sleeping scenario, and transmit the identification result to the AI ​​analysis module for analysis.

4. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: The AI ​​analysis module includes an embedded microprocessor and an algorithm based on a long short-term memory network, wherein: The algorithm is used to analyze historical data from weeks to months to generate personalized health thresholds and predict when heart failure will worsen; The AI ​​analysis module can dynamically adjust the threshold of abnormality detection according to the patient's activity status.

5. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: The biofeedback module includes a vibration motor and a Bluetooth audio output unit, wherein: The vibration motor is used to prompt the patient to adjust his breathing when shortness of breath or drop in blood oxygen is detected; The Bluetooth audio output unit is used to guide the patient to adjust the sleeping posture or breathing pattern through audio in a sleeping scene.

6. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: It also includes a data storage and communication module, which includes a flash memory chip and a low-power Bluetooth unit for storing historical data and transmitting monitoring results to an external device.

7. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: The wearable carrier is a chest patch or smart clothing, which is made of flexible circuits and breathable materials to ensure wearing comfort and long-term use stability.

8. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: The AI ​​analysis module can predict the risk of fluid overload by analyzing heart rate variability HRV and respiratory rate trends, and trigger the biofeedback module to intervene when the predicted result exceeds the personalized threshold.

9. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: Also included is a self-powered unit that powers the device through body heat or motion kinetic energy to reduce the need for external charging.

10. The intelligent device for monitoring the cardiopulmonary function of patients with heart failure according to claim 1, characterized in that: It also supports multi-user data management, which allows data to be stored in different accounts and shared remotely by connecting to mobile phone applications, making it suitable for home or community use scenarios.

Images