Wearable device for multi-modal physiological data acquisition and intelligent analysis

Abstract

The invention discloses a wearable device for multi-modal physiological data collection and intelligent analysis, and relates to the technical field of wearable devices. A data collection module collects original physiological data such as heart rate and blood pressure through various sensors; the preprocessing module performs normalization and downsampling on the data; the control analysis module extracts and fuses the feature vectors, and an evaluation result is obtained through a health evaluation model; the communication module transmits the data and the result to the intelligent terminal; and the power management module supplies power and performs power state detection and low electric quantity early warning. The problems that an existing wearable device is single in function, limited in data mode and insufficient in analysis capacity are solved, the physical condition of a user can be comprehensively and accurately monitored, the requirements of people for deeply understanding the health of the user and monitoring the health of the user in real time are met, meanwhile, stable operation of a device power source is guaranteed, and the practicability and reliability of the device are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wearable devices, and more specifically, to a wearable device for multi-modal physiological data acquisition and intelligent analysis. Background Art

[0002] At present, with the continuous improvement of people's health awareness and the rapid development of technology, wearable devices have been more and more widely used in the field of health monitoring. Although some wearable devices on the market can collect some physiological data, such as basic information like heart rate and steps, they have problems such as relatively single functions, limited data modalities collected, and difficulty in comprehensively and intelligently analyzing the collected multi-type data. They cannot comprehensively and accurately reflect the true physical condition of users and cannot meet people's growing needs for in-depth understanding and real-time monitoring of their own health conditions. Therefore, there is an urgent need for a wearable device that can simultaneously collect multi-modal physiological data and has a powerful intelligent analysis ability to fill this market gap. Summary of the Invention

[0003] In view of this, the present invention provides a wearable device for multi-modal physiological data acquisition and intelligent analysis to solve the problems in the background art.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] A wearable device for multi-modal physiological data acquisition and intelligent analysis, comprising: a data acquisition module, a data preprocessing module, a data control and analysis module, a communication module, and a power management module;

[0006] The data acquisition module collects the original physiological data signals of the user;

[0007] The data preprocessing module preprocesses the collected original physiological data signals to obtain preprocessed physiological data signals;

[0008] The data control and analysis module extracts features from the preprocessed physiological data signals to obtain multiple feature vectors, fuses the multiple feature vectors to obtain a comprehensive feature vector, and establishes a health assessment model to evaluate the comprehensive feature vector to obtain a health assessment result;

[0009] The communication module is connected to an external intelligent terminal and transmits the collected original physiological data signals and the health assessment result to the external intelligent terminal in real time;

[0010] The power management module is connected to the data control and analysis module and supplies power to each module.

[0011] Optionally, the data acquisition module includes a heart rate sensor, a blood pressure sensor, a blood oxygen saturation sensor, and a body temperature sensor; the original physiological data signals include heart rate data, blood pressure values, oxygen content in the blood, and body temperature of the human body.

[0012] Optionally, the preprocessing includes: first, normalizing using the standard deviation normalization method, and then downsampling through integer decomposition to regularize data with different sampling frequencies to a fixed scale.

[0013] Optionally, the data control and analysis module includes multiple feature extraction units, which respectively extract features from the preprocessed physiological data signals to obtain multiple feature vectors; each feature extraction unit is sequentially composed of 4 two-dimensional spatial convolution blocks and 1 two-dimensional adaptive max pooling layer, and each two-dimensional spatial convolution block contains 2 sub-modules, and each sub-module contains 1 two-dimensional spatial convolution, 1 batch normalization layer, and 1 Relu non-linear activation function.

[0014] Optionally, the data control and analysis module further includes a feature fusion unit, which performs weighted superposition fusion on the multiple feature vectors to form a comprehensive feature vector, specifically:

[0015]

[0016] where, W is the fused comprehensive feature vector; ω i is the weight coefficient of the feature vector Fi obtained by the i-th feature extraction unit for feature extraction; h is the smoothing coefficient; Φ is the error term; μ is the correction term.

[0017] Optionally, the data control and analysis module further includes a health assessment model establishment unit, which is used to establish a health assessment model to assign values and evaluate the comprehensive feature vector, and calculate and obtain the user's health status level, specifically:

[0018] Set several levels under each evaluation index, corresponding to different health degrees respectively, and assign scores to them, and use V = {A, B, C, D} to represent the index level corresponding to the comprehensive feature vector; where: V is the evaluation index, and A, B, C, D are different index levels;

[0019] Obtain the index level score V corresponding to each evaluation index according to the comprehensive feature vector i , and its expression is as follows:

[0020] V i = (x1, x2,..., xn)

[0021] where: x1, x2, xn are all the scores of the index level scores corresponding to the evaluation index;

[0022] The expression for calculating the health status score L of the user is as follows:

[0023]

[0024] According to the health status score of the user and the corresponding level of the score, the health status level of the user is obtained.

[0025] Optionally, the power management module includes a power status detection module. The process of comprehensively analyzing the temperature, power, and charge-discharge power of the power supply includes:

[0026] Calculate and obtain the temperature rise ratio coefficient K through the following formula;

[0027]

[0028] where t3 and t4 are the left and right endpoints of the preset target interval; ΔP is the average standby power; T(t) is the value of the power supply temperature changing with time; n is the preset exponent;

[0029] Compare the temperature rise ratio coefficient K with the preset threshold K t ;

[0030] If K < K t , it is determined that the power supply status is abnormal, and a corresponding power supply abnormality warning signal is emitted;

[0031] Otherwise, it is determined that the power supply status is good.

[0032] Optionally, the power management module further includes a low power warning unit, which is used to comprehensively analyze the temperature, power, and charge-discharge power of the power supply and perform low power warning. Specifically, it includes:

[0033] Calculate and obtain the estimated average power consumption P of the power supply discharge through the following formula dc ;

[0034]

[0035] If P dc > P chg , then calculate and obtain the estimated power consumption time t through the formula ; pre ;

[0036] If t pre - t sus < m, a low power warning is issued;

[0037] where t5 and t6 are the left and right endpoints of the preset unit interval [t5, t6] respectively; P chg is the standard charging power of the power supply; P ε is the preset error coefficient; Q restis the remaining power of the power supply; ρ is a preset adjustment coefficient, m is a preset safety difference, and m > 0.

[0038] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a wearable device for multi-modal physiological data collection and intelligent analysis. The data collection module collects original physiological data such as heart rate and blood pressure through multiple sensors; the preprocessing module normalizes and down-samples the data; the control and analysis module extracts and fuses feature vectors, and obtains an evaluation result through a health assessment model; the communication module transmits the data and results to the intelligent terminal; the power management module supplies power and performs power status detection and low power warning. The present invention solves the problems of single function, limited data modalities and insufficient analysis ability of existing wearable devices, can comprehensively and accurately monitor the physical condition of users, meet people's needs for in-depth understanding and real-time monitoring of their own health, and at the same time ensure the stable operation of the device power supply, improving the practicability and reliability of the device. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0040] Figure 1 is the system structure diagram provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments 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. 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.

[0042] An embodiment of the present invention discloses a wearable device for multi-modal physiological data collection and intelligent analysis, as Figure 1 shown, including: a data collection module, a data preprocessing module, a data control and analysis module, a communication module, and a power management module;

[0043] The data collection module collects the original physiological data signals of the user;

[0044] The data preprocessing module preprocesses the collected original physiological data signals to obtain preprocessed physiological data signals;

[0045] A data control and analysis module extracts features from the preprocessed physiological data signals to obtain multiple feature vectors, fuses the multiple feature vectors to obtain a comprehensive feature vector, and establishes a health assessment model to evaluate the comprehensive feature vector to obtain a health assessment result;

[0046] A communication module connects to an external intelligent terminal and transmits the collected original physiological data signals and health assessment results to the external intelligent terminal in real time;

[0047] A power management module establishes a connection with the data control and analysis module and supplies power to each module.

[0048] In a specific embodiment, the data acquisition module includes a heart rate sensor, a blood pressure sensor, a blood oxygen saturation sensor, and a body temperature sensor; the original physiological data signals include heart rate data, blood pressure values, oxygen content in the blood, and body temperature. In terms of the hardware assembly of the device, the data acquisition module carefully selects high-precision heart rate sensors, blood pressure sensors, blood oxygen saturation sensors, and body temperature sensors and arranges them reasonably at the corresponding positions of the wearable device to ensure that the original physiological data signals of the user can be accurately and stably collected. These sensors are tightly connected to the inside of the device. Once the device is started and worn on the user, they immediately start working to obtain information such as heart rate data, blood pressure values, oxygen content in the blood, and body temperature in real time.

[0049] In a specific embodiment, the preprocessing includes: first, normalization is performed using the standard deviation normalization method, and then downsampling is performed through integer decomposition to regularize data with different sampling frequencies to a fixed scale. The first step of data preprocessing is data normalization. Since the time-series physiological data is collected from different organs and by different acquisition devices, the signal-to-noise ratio in the original data will be different. For example, for the data collected by non-contact devices, due to the differences in the body postures of the patients (sideways and upright), the amplitude of the collected signals will fluctuate greatly. For the same disease pattern, there are different manifestations under multiple amplitude scales. The normalization method in this embodiment can adopt standard deviation normalization. The time-series physiological data collected by different types of devices usually has different sampling rates. In order to be able to perform unified modeling, such data usually requires a downsampling operation to regularize data with different sampling frequencies to a fixed scale, such as downsampling data with different sampling frequencies to 50 Hz. The downsampling in this embodiment can be implemented by methods such as integer decomposition.

[0050] In a specific embodiment, the data control and analysis module includes multiple feature extraction units. The multiple feature extraction units respectively perform feature extraction on the preprocessed physiological data signals to obtain multiple feature vectors. Each feature extraction unit is sequentially composed of 4 two-dimensional spatial convolution blocks and 1 two-dimensional adaptive max pooling layer. Each two-dimensional spatial convolution block contains 2 sub-modules, and each sub-module contains 1 two-dimensional spatial convolution, 1 batch normalization layer, and 1 Relu non-linear activation function.

[0051] In specific implementation, before the two-dimensional spatial convolution blocks, 1 two-dimensional convolution with a convolution kernel of 7×7, edge padding of 3, and a stride of 2, 1 batch normalization, and 1 Relu non-linear activation function operation are also sequentially adopted. Each feature extraction network unit shares parameters. The two-dimensional spatial convolution parameters in the first sub-module are: convolution kernel size 3×3, edge padding of 1 for both, and a stride of 2 for both. The two-dimensional spatial convolution parameters in the second sub-module are: convolution kernel size 3×3, edge padding of 1 for both, and a stride of 1 for both.

[0052] In a specific embodiment, the data control and analysis module further includes a feature fusion unit that performs weighted superposition fusion on the multiple feature vectors to form a comprehensive feature vector. Specifically, it is expressed as:

[0053]

[0054] where, W is the fused comprehensive feature vector; ω i is the weight coefficient of the feature vector Fi obtained by the i-th feature extraction unit for feature extraction; h is the smoothing coefficient; Φ is the error term; μ is the correction term.

[0055] In a specific embodiment, the data control and analysis module further includes a health assessment model establishment unit, which is used to establish a health assessment model to evaluate the comprehensive feature vector and calculate the user's health status level. Specifically:

[0056] Several levels are set under each evaluation index, corresponding to different health degrees, and scores are assigned to them. Use V = {A, B, C, D} to represent the index level corresponding to the comprehensive feature vector. Among them: V is the evaluation index, and A, B, C, D are different index levels;

[0057] According to the comprehensive feature vector, obtain the index level score V corresponding to each evaluation index i , and its expression is as follows:

[0058] V i =(x1, x2,..., xn)

[0059] where: x1, x2, xn are all the scores of the index level scores corresponding to the evaluation index;

[0060] The expression for calculating the user's health status score L is as follows:

[0061]

[0062] According to the user's health status score and the corresponding level, the user's health status level is obtained, specifically referring to Table 1.

[0063] Table 1 Health Status Level Table

[0064] Health status level Level 1 Level 2 Level 3 Level 4 Score 90-100 80-90 60-80 0-60

[0065] In a specific embodiment, the power management module includes a power status detection module. The process of comprehensively analyzing the temperature, power, and charge-discharge power of the power supply includes:

[0066] The heating-up ratio coefficient K is calculated and obtained through the following formula;

[0067]

[0068] where t3 and t4 are the left and right endpoints of the preset target interval; ΔP is the average standby power; T(t) is the value of the power supply temperature changing with time; n is the preset exponent;

[0069] The heating-up ratio coefficient K is compared with the preset threshold K t ;

[0070] If K < K t , it is determined that the power supply status is abnormal, and the corresponding power supply abnormality warning signal is emitted;

[0071] Otherwise, it is determined that the power supply status is good.

[0072] It should be noted that under normal power supply conditions, there is a non-linear relationship between power and temperature, but an exponential relationship. The preset exponent n in the formula can be specifically obtained according to the experimental data of the standard power supply.

[0073] In a specific embodiment, the power management module further includes a low power warning unit, which is used to comprehensively analyze the temperature, power, and charge-discharge power of the power supply for low power warning, specifically including:

[0074] The estimated average power consumption of the power supply discharge P is calculated and obtained through the following formula dc ;

[0075]

[0076] If P dc > P chg , then the estimated power consumption time t is calculated and obtained through the formula ​pre ;

[0077] If t pre -t sus < m, a low - power warning is issued;

[0078] Among them, t5 and t6 are the left and right endpoints of the preset unit interval [t5, t6] respectively; P chg is the standard charging power of the power supply; P ε is the preset error coefficient; Q rest is the remaining power of the power supply; ρ is the preset adjustment coefficient, m is the preset safety difference, and m > 0. This safety difference is to give the user enough time to charge or take other measures before the power is about to run out, ensuring that the normal use of the device will not be interrupted due to power problems.

[0079] In terms of communication, the communication module adopts advanced wireless communication technologies such as Bluetooth or Wi - Fi to establish a stable connection with external intelligent terminals (such as mobile phones, tablets, etc.). Once the data control and analysis module obtains the original physiological data signal and the health assessment result, the communication module immediately transmits these data to the external intelligent terminal in real time. On the intelligent terminal, the user can view the detailed physiological data and health assessment report through a specially developed application program. The application program can also further visually display the data, for example, presenting the change trends of data such as heart rate and blood pressure in the form of charts, facilitating the user to intuitively understand their physical condition.

[0080] In this specification, each embodiment is described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0081] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A wearable device for multimodal physiological data collection and intelligent analysis, characterized in that: include: Data acquisition module, data preprocessing module, data control and analysis module, communication module, power management module; The data acquisition module collects the original physiological data signals of the user; The data preprocessing module preprocesses the collected original physiological data signal to obtain a preprocessed physiological data signal; The data control and analysis module extracts features from the preprocessed physiological data signals to obtain multiple feature vectors, fuses the multiple feature vectors to obtain a comprehensive feature vector, establishes a health assessment model to evaluate the comprehensive feature vector, and obtains a health assessment result; The communication module is connected to an external intelligent terminal to transmit the collected original physiological data signals and health assessment results to the external intelligent terminal in real time; The power management module establishes a connection with the data control and analysis module and supplies power to each module.

2. A wearable device for multimodal physiological data collection and intelligent analysis according to claim 1, characterized in that: The data acquisition module includes a heart rate sensor, a blood pressure sensor, a blood oxygen saturation sensor, and a body temperature sensor; the original physiological data signal includes heart rate data, blood pressure value, oxygen content in the blood, and human body temperature.

3. A wearable device for multimodal physiological data collection and intelligent analysis according to claim 1, characterized in that: The preprocessing includes: firstly, normalizing by using the standard deviation normalization method, and then downsampling by integer decomposition to regularize the data of different sampling frequencies to a fixed scale.

4. A wearable device for multimodal physiological data collection and intelligent analysis according to claim 1, characterized in that: The data control and analysis module includes multiple feature extraction units, which respectively extract features from the preprocessed physiological data signals to obtain multiple feature vectors; each feature extraction unit is composed of 4 two-dimensional spatial convolution blocks and 1 two-dimensional adaptive maximum pooling layer in sequence, each two-dimensional spatial convolution block contains 2 sub-modules, and each sub-module contains 1 two-dimensional spatial convolution, 1 batch normalization layer, and 1 Relu nonlinear activation function.

5. A wearable device for multimodal physiological data collection and intelligent analysis according to claim 4, characterized in that: The data control and analysis module further includes a feature fusion unit, which performs weighted superposition and fusion on the multiple feature vectors to form a comprehensive feature vector, which is specifically manifested as follows: Among them, W is the integrated feature vector after fusion; ω i is the weight coefficient of the feature vector Fi obtained by feature extraction by the i-th feature extraction unit; h is the smoothing coefficient; Φ is the error term; μ is the correction term.

6. A wearable device for multimodal physiological data collection and intelligent analysis according to claim 1, characterized in that: The data control and analysis module also includes a health assessment model establishment unit, which is used to establish a health assessment model to perform value assignment and evaluation on the comprehensive feature vector, and calculate the user's health status level, specifically: Several levels are set up under each evaluation index, corresponding to different health levels, and scores are assigned to them. V = {A, B, C, D} is used to represent the index level corresponding to the comprehensive feature vector; where: V is the evaluation index, and A, B, C, and D are different index levels; According to the comprehensive feature vector, the index grade score V corresponding to each evaluation index is obtained i , its expression is: V i =(x1,x2,…,xn) Among them: x1, x2, xn are the scores of the indicator level corresponding to the evaluation indicators; The expression for calculating the user's health status score L is as follows: The user's health status level is obtained based on the user's health status score and the level corresponding to the score.

7. The wearable device for multimodal physiological data collection and intelligent analysis according to claim 1, characterized in that: The power management module includes a power status detection module, and the process of comprehensively analyzing the temperature, power, and charge and discharge power of the power supply includes: The temperature rise proportional coefficient K is obtained by calculating the following formula: Wherein, t3 and t4 are the left and right endpoints of the preset target interval; ΔP is the average standby power; T(t) is the value of the power supply temperature changing with time; n is the preset index; The temperature rise proportional coefficient K and the preset threshold K t Make a comparison; If K <K t , the power supply status is judged to be abnormal, and the corresponding power supply abnormality warning signal is issued; Otherwise, the power supply is considered good.

8. The wearable device for multimodal physiological data collection and intelligent analysis according to claim 1, characterized in that: The power management module also includes a low power warning unit, which is used to perform a comprehensive analysis of the temperature, power and charge and discharge power of the power supply and issue a low power warning, specifically including: The estimated average power P of the power supply discharge can be calculated by the following formula dc ; If P dc >P chg , then by the formula Calculate and obtain the estimated power consumption time t pre ; If t pre -t sus < m, then a low battery warning is issued; Wherein, t5 and t6 are the left and right endpoints of the preset unit interval [t5, t6] respectively; chg is the standard charging power of the power source; P ε is the preset error coefficient; Q rest is the remaining power of the power supply; ρ is the preset adjustment coefficient, m is the preset safety difference, and m>0.

Images