A multi-person physiological information detection-separation method and system

Abstract

The application discloses a kind of multi-human physiological information detection-separation method and system, comprising: the mixed signal of human motion is collected;All signals in the collected signal are screened, filtered, and the gait signal in the mixed signal is separated;The signal after the gait signal is deducted is time-domain segmentation, the data of the segmented mixed signal is used as the input of neural network model, and the model is trained by adjusting parameters, the indicators of different parameter models are compared, and the optimal model is saved;Respiratory signal and heart rate signal are separated using the model after training, and the frequency of three signals is calculated.The application is relatively simple on the device and has less restraint on the human body, and the sensor also has the characteristics of small size, light weight and low power consumption;The complexity of the model training of the application is lower, the application reduces the loss of information, increases the robustness of the model, can better avoid overfitting, and improves the generalization ability of the network.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of human physiological signal separation, and in particular to a multi-human physiological information detection-separation method and system. BACKGROUND

[0002] Heart rate is an important parameter for observing the health status of vital signs, and in medicine, the detection of heart rate can be performed by electrocardiograph and echocardiography devices, etc. However, these measuring devices are often large in size and have certain binding to the person when used, and thus have poor convenience. On the other hand, smart wristbands and wristbands, compared to medical devices, have very good use convenience, but due to the large movement displacement of the bones and muscles of the wrist joint, the sensors placed in such devices are difficult to stably obtain information at the same position, and thus the distortion degree of the physiological signals output by the devices is high. Compared with the above two types of devices, a smart chest band is a device that can simultaneously consider operation convenience and data accuracy, and the sensor placed in the smart chest band can be a temperature sensor for measuring body temperature, a conductive electrode for measuring electrocardiogram parameters, or a photoelectric or acceleration sensor for measuring heart rate and heart rate variability, and thus the smart chest band has good application prospects.

[0003] The smart chest band with an acceleration sensor has the characteristics of low cost, small size and low power consumption, but due to its high measurement sensitivity, the physiological signals obtained by such a smart chest band often also mix in respiratory rhythm and gait information, and how to obtain accurate heart rate data through a suitable signal processing method is a challenging task. SUMMARY

[0004] This section is intended to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, the abstract and the title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0005] In view of the above existing problems, the present application is proposed.

[0006] Therefore, the technical problem solved by the present application is how to obtain accurate heart rate data through a suitable signal processing method.

[0007] To solve the above technical problems, the present application provides the following technical solutions: collecting mixed signals during human motion; filtering and processing all signals in the collected signals, and separating gait signals in the mixed signals; performing time domain segmentation on the signals after the gait signals are deducted, taking the data of the segmented mixed signals as the input of the neural network model, and saving the optimal model by adjusting parameters and comparing the indicators of different parameter models; using the trained model to separate the respiratory signal and the heart rate signal, and calculating the frequencies of the three signals.

[0008] As a preferred scheme of the multi-human physiological information detection-separation method of the present application, the gait signal in the mixed signal is extracted using a band-pass filter.

[0009] As a preferred scheme of the multi-human physiological information detection-separation method of the present application, all collected data are deleted for bad points, and the respiratory and heart rate signals are filtered; the filtering process uses a digital filter: the weak heart rate signal and body noise in the respiratory signal are eliminated by a Butterworth low-pass filter; the baseline drift of the heart rate signal is eliminated by a Butterworth high-pass filter.

[0010] As a preferred scheme of the multi-human physiological information detection-separation method of the present application, the Conv-Tasnet deep learning network model is used to separate the signals, including: taking the data after data preprocessing as the training data of the model, and performing time domain segmentation on it; adjusting the parameters of the model, setting the learning rate of the model, calculating the loss value by forward propagation, performing back propagation, and modifying the weight; if the loss value reaches the preset indicator, the model is saved, otherwise it is judged whether the maximum iteration number is reached, if the maximum iteration number is reached, the model is saved, otherwise the parameters are continuously adjusted.

[0011] As a preferred scheme of the multi-human physiological information detection-separation method of the present application, the Conv-Tasnet deep learning network model adds Bahdanau attention mechanism in the separator, and adds a random removal unit dropout after the separator; the gradient descent method used by the Conv-Tasnet deep learning network model is AdamW+SAM.

[0012] As a preferred scheme of the multi-human physiological information detection-separation method of the present application, the selection standard of the optimal separation model is the optimal model obtained by comparing the performance of different parameter trained models, including two indicators: signal-to-noise ratio improvement and maximum scale invariant signal-to-noise ratio; the model with optimal signal-to-noise ratio improvement or maximum scale invariant signal-to-noise ratio is selected as the signal separation model.

[0013] As a preferred scheme of the multi-human physiological information detection-separation method, for the heart rate signal, an interval maximum value algorithm is used to obtain each peak value of the heart rate signal, and a heart rate is calculated through a signal frequency calculation formula; for the respiration signal and the gait signal, high-frequency signals are filtered out through a low-pass filter, a wave peak extraction method is used to find the number of wave peaks in an interval, and a respiration frequency and a step frequency are calculated through the signal frequency calculation formula.

[0014] As a preferred scheme of the multi-human physiological information detection-separation method, the signal frequency calculation formula is:

[0015]

[0016] wherein, f s is a sampling frequency, X start is a position of the first peak value, X end is a position of the last peak value, and n is the number of the peak values in the interval.

[0017] To solve the above technical problems, the application further provides a multi-human physiological information detection-separation system, comprising: a signal collector comprising an acceleration sensor, a microcontroller module, an analog-to-digital conversion unit and a wireless transmission module, which are connected through a wired or wireless mode, and are used to collect mixed signals during human motion; a data preprocessing subsystem connected with the signal collector, comprising a wireless receiving module, a screening and filtering module and a gait separation module, and being used to preprocess the collected signals; and a signal separation subsystem connected with the data preprocessing subsystem, comprising a model training module, a signal separation module and a frequency calculation module, and being used to construct a model and separate signals.

[0018] The application has the advantages that: the application is simple in equipment and has less restraint on the human body, and the sensor has the characteristics of small size, light weight and low power consumption; the model training of the application has low complexity and small size; the application reduces information loss, increases the robustness of the model, can better avoid overfitting, and improves the generalization ability of the network. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor. Among them:

[0020] Figure 1A basic flowchart of a multi-person physiological information detection-separation method and system provided by an embodiment of the present application is shown in the figure;

[0021] Figure 2 A signal analysis algorithm flowchart of a multi-person physiological information detection-separation method and system provided by an embodiment of the present application is shown in the figure;

[0022] Figure 3 A signal collector wearing diagram of a multi-person physiological information detection-separation method and system provided by an embodiment of the present application is shown in the figure;

[0023] Figure 4 A data preprocessing subsystem working flowchart of a multi-person physiological information detection-separation method and system provided by an embodiment of the present application is shown in the figure;

[0024] Figure 5 A model training flowchart of a multi-person physiological information detection-separation method and system provided by an embodiment of the present application is shown in the figure;

[0025] Figure 6 A signal diagram of a signal separation flowchart of a multi-person physiological information detection-separation method and system provided by an embodiment of the present application is shown in the figure;

[0026] Figure 7 A module structure diagram of a multi-person physiological information detection-separation method and system provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0027] In order to make the above objectives, features and advantages of the present application more apparent, specific embodiments of the present application will be described in detail below with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work should fall within the protection scope of the present application.

[0028] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be practiced without the specific details, which are not described in the present application, and it can be apparent to those skilled in the art that the present application can be implemented in other different ways. Therefore, the present application is not limited to the specific embodiments disclosed below.

[0029] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. The "in one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or selective embodiment which excludes other embodiments.

[0030] The present application is described in detail in conjunction with the schematic diagram, and in the detailed description of the embodiments of the present application, the cross-sectional view of the device structure is partially enlarged without the general proportion for the convenience of illustration, and the schematic diagram is only an example, which should not limit the scope of protection of the present application herein. In addition, the three-dimensional spatial dimensions of length, width and depth should be included in actual production.

[0031] Meanwhile, in the description of the present application, it should be noted that the terms "upper, lower, inner and outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first, second or third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0032] Unless otherwise specifically defined and limited, the terms "mounting, connecting, connecting" in the present application should be broadly understood, for example: it can be fixedly connected, detachably connected or integrally connected; it can also be mechanically connected, electrically connected or directly connected, it can also be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0033] Example 1

[0034] Reference Figures 1-6 For an embodiment of the present application, a multi-physiological information detection-separation method is provided, comprising:

[0035] S1: Collecting the mixed signal during human motion.

[0036] It should be noted that the mixed signal during human motion is collected by a signal collector, wherein, as shown in Figure 3 , the signal collector is fixed on the elastic chest strap, and in the design of the experiment to collect physiological signals, the signal collector is placed at the left chest of the human body, and the collected signal is the heart rate, respiration, gait and mixed signal of the three signals of the subject.

[0037] S2: Separate the gait signal in the mixed signal, and filter and process all signals in the mixed signal.

[0038] It should be noted that, as Figure 4As shown, all collected data are subjected to bad point deletion and filtering of respiratory and heart rate signals, wherein the filtering is performed by a digital filter method: the weak heart rate signal and body noise in the respiratory signal are eliminated by a Butterworth low-pass filter; the baseline drift of the heart rate signal is eliminated by a Butterworth high-pass filter, and the gait signal in the mixed signal is extracted by a band-pass filter.

[0039] S3: The signal after the gait signal is deducted is subjected to time domain segmentation, the data of the segmented mixed signal is used as the input of the neural network model, and the optimal model is saved by adjusting the parameters and comparing the indicators of different parameter models.

[0040] It should be noted that, preferably, as shown in Figure 2 , 5 The neural network model adopts a Conv-Tasnet deep learning network model, and the main structure includes an encoder, a separator and a decoder. The training and generation process includes: the data processed by the data preprocessing subsystem is used as the original training data, which is segmented in time domain, then the parameters of the model are adjusted, the learning rate of the model is set, the loss value is calculated by forward propagation, the weight is modified by back propagation, if the loss value reaches the expected indicator, the model is saved, otherwise continue to judge whether the maximum iteration number is reached, if the maximum iteration number is reached, the model is saved, otherwise continue to adjust the parameters.

[0041] Further, the Conv-Tasnet deep learning network model described above adds Bahdanau attention mechanism in the separator to reduce information loss. To avoid overfitting and increase the robustness of the model, a random removal unit dropout is added after the separator. The gradient descent method adopted by the Conv-Tasnet deep learning network model is AdamW+SAM. AdamW is an improved algorithm based on Adam+L2 regularization. Specifically, it optimizes the defect that Adam punishes the parameters with larger weights but not the parameters with smaller weights. Sharpness-aware minimization (SAM) reduces the loss value and loss sharpness, finds parameters with uniform loss value in its field, and improves the generalization ability of the network.

[0042] Further, the selection standard of the optimal separation model is the optimal model obtained by comparing the performance of models trained with different parameters, including two indicators: signal-to-noise ratio improvement (SDRi) and maximum scale invariant signal-to-noise ratio (SI-SNR). The model with optimal signal-to-noise ratio improvement or maximum scale invariant signal-to-noise ratio is selected as the signal separation model.

[0043] S4: The model after training is used to separate the respiratory signal and the heart rate signal, and the frequencies of the three signals are calculated.

[0044] It should be noted that for the heart rate signal, the interval maximum value algorithm is adopted to obtain each peak value of the heart rate signal, and the heart rate is calculated through the signal frequency calculation formula; for the respiratory signal and the gait signal, high-frequency signals are filtered out through a low-pass filter, the number of wave peaks in the interval is found through the wave peak extraction method, and the respiratory frequency and the step frequency are calculated through the signal frequency calculation formula.

[0045] Preferably, the signal frequency calculation formula is:

[0046]

[0047] Wherein, f s is the sampling frequency, X start is the position of the first peak value, X end is the position of the last peak value, and n is the number of peak values in the interval.

[0048] The physiological signal acquisition device of the present application is composed of an elastic chest strap and an acceleration sensor, which is relatively simple on the device and has less binding to the human body, and the sensor also has the characteristics of small size, light weight and low power consumption; the method of the present application separates the respiratory, heart rate and gait signals under dynamic conditions, extracts the gait signal first, then separates the respiratory and heart rate signals through the separation model, reduces the complexity of model training and reduces the model size; the deep learning model of the present application is Conv-Tasnet, Bahdanau attention mechanism is added in front of the decoder, which reduces the loss of information, dropout layer is added behind the separator, which increases the robustness of the model and can better avoid overfitting, and the gradient descent method is AdamW+SAM, which can find uniform loss value parameters in its field and improve the generalization ability of the network.

[0049] The technical effects used in the method are verified and explained, and the present embodiment uses the method for testing, and the scientific demonstration means are used to verify the real effect of the method.

[0050] The experimenter needs to collect the heart rate and respiratory signal under static state, on the treadmill, the speed is 2.5km / h, and the speed is increased step by step, each time by 0.5km / h, and the respiratory signal, heart rate signal, gait signal and mixed signal of the three signals of the experimenter are collected, the data is sent to the data preprocessing subsystem by the signal collector, the data is screened and filtered, and then the maximum scale invariant signal-to-noise ratio optimal model is obtained through the training module.

[0051] Figure 1 is a multi-human signal separation workflow diagram, such as Figure 6Signal 'a' shown represents the raw mixed signal acquired by the signal acquisition unit. The gait signal is obtained by filtering the raw acceleration mixed signal through a digital bandpass filter. Figure 6 The b signal is shown; then the mixed signal after subtracting the gait signal is obtained, as shown. Figure 6 The signal shown as c is transmitted to the signal separation module, where the heart rate and respiratory signals are separated using the Conv-Tasnet model. Finally, the frequencies of the three signals are calculated using the following methods and formulas: For the heart rate signal, an interval maximum value algorithm is used to find each peak value of the heart rate signal, and then the heart rate is calculated using the formula; for the respiratory and gait signals, a low-pass filter is used to filter out high-frequency signals, and a peak extraction method is used to find the number of peaks within an interval, and then the respiratory rate and gait rate are calculated using the formulas. The following formulas are used to calculate the frequencies of the three signals:

[0052]

[0053] Among them, f s It is the sampling frequency, X start It is the location of the first peak, X end It is the location of the last peak, and n is the number of peaks in this interval.

[0054] The experimental results above show that the acceleration-based human multi-physiological signal separation method and system proposed in this invention has a relatively accurate physiological signal acquisition process, is relatively simple to implement, is less restrictive and convenient to wearable devices, and provides more accurate performance evaluation of signal separation results.

[0055] Example 2

[0056] Reference Figure 7 In another embodiment of the present invention, which differs from the first embodiment, a multi-person physiological information detection-separation system is provided. The aforementioned multi-person physiological information detection-separation method is implemented based on this system, and specifically includes:

[0057] The signal acquisition unit includes an accelerometer, a microcontroller module, an analog-to-digital converter, and a wireless transmission module. It is connected via wired or wireless means and is used to acquire mixed signals during human movement. Preferably, the accelerometer is used to acquire human physiological information. The microcontroller module is used to filter out high-frequency noise and control the transmission of signals. The analog-to-digital converter converts analog signals into digital signals. The wireless transmission module sends digital signals to the data preprocessing subsystem.

[0058] The data preprocessing subsystem, connected with the signal collector, comprises a wireless receiving module, a screening and filtering module, and a gait separation module, and is used for preprocessing the collected signals. Preferably, the wireless receiving module is used for receiving the data transmitted by the signal collector; the screening and filtering module is used for screening bad points and filtering out interference noise in the collected multiple physiological signals; and the gait separation module is used for separating the gait signal from the mixed signal of the heart rate, the respiration and the gait.

[0059] The signal separation subsystem, connected with the data preprocessing subsystem, comprises a model training module, a signal separation module and a frequency calculation module, and is used for constructing a model and separating signals. Preferably, the model training module is used for obtaining the best signal separation model through training; the signal separation module is used for separating the heart rate and the respiration signal from the mixed signal from which the gait signal is deducted; and the frequency calculation module is used for calculating the step frequency, the heart rate and the respiration frequency through the separated signals.

[0060] It should be appreciated that embodiments of the present application can be realized or implemented by computer hardware, a combination of hardware and software, or through computer instructions stored in a non-transitory computer readable storage medium. The method can be implemented in a computer program using standard programming techniques, including a non-transitory computer readable storage medium configured with a computer program, wherein the storage medium thus configured makes the computer operate in a specific and predefined manner according to the method described in the specific embodiments and the accompanying drawings. Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with a computer system. However, if necessary, the program can be implemented in assembly or machine language. In any case, the language can be a compiled or interpreted language. In addition, the program can be run on a programmed special integrated circuit for this purpose.

[0061] In addition, the operations of the processes described herein can be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context. The processes described herein (or variations and / or combinations thereof) can be performed under the control of one or more computer systems configured with executable instructions (e.g., computer programs, one or more computer programs, or one or more applications) to perform the processes, and can be implemented as code (e.g., executable instructions, one or more computer programs, or one or more applications) executing collectively on one or more processors, by hardware, or combinations thereof. The computer programs include a plurality of instructions executable by one or more processors.

[0062] Further, the methods can be implemented in any type of computing platform operatively coupled to a suitable computing platform, including but not limited to a personal computer, a mini-computer, a mainframe, a workstation, a network or distributed computing environment, a stand-alone or integrated computer platform, or in communication with a charged particle tool or other imaging device, and the like. Aspects of the present application can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated to the computing platform, such as a hard disk, an optical read and / or write storage medium, RAM, ROM, and the like, such that it can be read by a programmable computer to configure and operate the computer to perform the processes described herein when the storage medium or device is read by the computer. In addition, the machine-readable code, or portions thereof, can be transmitted over a wired or wireless network. The present application, as described herein, includes these and other non-transitory computer readable storage media of different types when such media include instructions or programs implementing the steps described above in connection with a microprocessor or other data processor. The present application also includes the computer itself when programmed according to the methods and techniques described herein. The computer program can be applied to input data to perform the functions described herein and thereby transform the input data into output data that can be stored to non-volatile memory. The output information can further be applied to one or more output devices such as a display. In a preferred embodiment of the present application, the transformed data represents a physical and tangible object, including a particular visual depiction of a physical and tangible object produced on a display.

[0063] As used in this application, the terms "component," "module," "system" and the like are intended to refer to a computer-related entity, either hardware, firmware, a combination of hardware and software, software, or software in execution. For example, a component can be, but is not limited to being, a process running on a processor, a processor, an object, an executable, a thread of execution, a program, and / or a computer. By way of illustration, both an application running on a computing device and the computing device can be a component. One or more components can reside within a process and / or thread of execution and a component can be localized, partially and / or entirely, in one computer or distributed between two or more computers. In addition, these components can execute from various computer readable media having various data structures stored thereon. The components can communicate by way of local and / or remote processes such as in accordance with a signal having one or more data packets (e.g., data from one component interacting with another component in a local system, distributed system, and / or across a network such as the Internet with other systems via the signal).

[0064] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A multi-person physiological information detecting-separating method, characterized by, include: Collect mixed signals during human movement; All signals in the acquired signals are screened and filtered, and the gait signal in the mixed signal is separated. The gait signal is subtracted and then segmented in the time domain. The segmented mixed signal data is used as the input to the neural network model. The optimal model is saved by adjusting the parameters and comparing the performance of different parameter models. The respiratory signal and heart rate signal were separated using the trained model, and the frequencies of the three signals were calculated. Gait signals are extracted from the mixed signal using a bandpass filter; Signal separation is performed using the Conv-Tasnet deep learning network model, including: The preprocessed data is used as the training data for the model, and it is then split in the time domain. Adjust the model parameters, set the model's learning rate, calculate the loss value through forward propagation, perform backpropagation, and modify the weights; If the loss value reaches the preset index, the model is saved; otherwise, it continues to check whether the maximum number of iterations has been reached. If the maximum number of iterations has been reached, the model is saved; otherwise, the parameters are adjusted. The Conv-Tasnet deep learning network model incorporates a Bahdanau attention mechanism in the separator and a dropout unit after the separator; the gradient descent method used in the Conv-Tasnet deep learning network model is AdamW+SAM.

2. The multi-physiological information detecting-separating method according to claim 1, wherein, Also includes: All collected data were deleting bad pixels, and the respiratory and heart rate signals were filtered. The filtering process employs digital filters: a Butterworth low-pass filter eliminates weak heart rate signals and body noise in the respiratory signal; and a Butterworth high-pass filter eliminates baseline drift in the heart rate signal.

3. The multi-physiological information detecting-separating method according to claim 2, wherein, The selection criteria for the optimal separation model are the best model obtained by comparing the performance of models trained with different parameters, including two indicators: signal-to-noise ratio improvement and maximizing scale-invariant signal-to-noise ratio; the model with the best signal-to-noise ratio improvement or maximizing scale-invariant signal-to-noise ratio is selected as the signal separation model.

4. The method for detecting and separating multiple human physiological information as described in claim 3, characterized in that, Also includes: For the heart rate signal, the interval maximum value algorithm is used to obtain each peak value of the heart rate signal, and the heart rate is calculated using the signal frequency calculation formula; For the respiratory and gait signals, high-frequency signals are filtered out by a low-pass filter, the number of peaks in the interval is found by the peak extraction method, and the respiratory frequency and gait frequency are calculated by the signal frequency calculation formula.

5. The method for detecting and separating physiological information from multiple individuals as described in claim 4, characterized in that, The formula for calculating the signal frequency is: in, It is the sampling frequency. It is the location of the first peak. It is the location of the last peak. It represents the number of peak values ​​in this interval.

6. A multi-person physiological information detection-separation system, employing the multi-person physiological information detection-separation method as described in any one of claims 1 to 5, characterized in that, include: The signal acquisition device includes an accelerometer, a microcontroller module, an analog-to-digital converter, and a wireless transmission module. It is connected wirelessly and used to acquire mixed signals during human movement. The data preprocessing subsystem, connected to the signal acquisition unit, includes a wireless receiving module, a screening and filtering module, and a gait separation module, used to preprocess the acquired signals; The signal separation subsystem, connected to the data preprocessing subsystem, includes a model training module, a signal separation module, and a frequency calculation module, used to construct the model and separate the signal.

Images