Fall early warning method and system based on multi-electrophysiology

Abstract

The invention discloses a tumble early warning method and system based on multi-electrophysiology, and is applied to the technical field of gait protection, and the method comprises the steps: carrying out the feature extraction of a multi-modal electrophysiology signal, and obtaining a multi-modal electrophysiology feature; respectively carrying out feature extraction on the electromyographic signal and the electroencephalogram signal to obtain a first single-point electrophysiological feature and a second single-point electrophysiological feature; evaluating the muscle state according to the electromyographic signal to obtain a first evaluation result; evaluating the brain function state according to the electroencephalogram signal to obtain a second evaluation result; calculating reaction time of the electromyographic signal and the electroencephalogram signal to obtain a third evaluation result; inputting the obtained individual information and the evaluation result into a fall risk evaluation model, and outputting a risk coefficient; inputting the multi-modal electrophysiological features, the single-point electrophysiological features and the risk coefficient into a tumble early warning model for training; and predicting the multi-mode electrophysiological signal of the target user based on the trained tumble early warning model, and outputting a tumble early warning result.

Description

Technical Field

[0001] The present invention relates to the technical field of gait protection, and in particular to a multi-electrophysiological based fall warning method and system. Background Art

[0002] Functional degeneration and lower limb dysfunction in the elderly will greatly increase the risk of falls in the elderly. Therefore, the research and development of fall protection for the elderly and people with lower limb dysfunction has become an urgent need in clinical and community settings.

[0003] There are three main types of fall protection devices at present: walking aids, follow-up fall protection devices, and wearable fall protection airbags. However, no matter which method is used, there should be an accurate and timely prediction to work normally and effectively. The current fall detection methods mainly include: millimeter wave detection method, video detection method, and audio detection method. These methods are prone to misjudgment due to the single information. At the same time, due to the lag of this information, the warning is not timely, and it is difficult to detect on the body, and it is not convenient to link with the protection device.

[0004] In order to overcome these defects, the present application proposes a fall warning method and system based on multi-electrophysiology. Summary of the invention

[0005] The purpose of this application is to provide a multi-electrophysiological based fall warning method and system, aiming to solve the problem that the existing fall warning is not timely enough and prone to misjudgment.

[0006] To achieve the above objectives, this application provides the following technical solutions: In a first aspect, the present application provides a multi-electrophysiological fall warning method, comprising: Acquiring multimodal electrophysiological signals, wherein the multimodal electrophysiological signals include electromyographic signals and electroencephalographic signals; Performing feature extraction on the multimodal electrophysiological signal to obtain a multimodal electrophysiological feature; performing feature extraction on the electromyographic signal and the electroencephalographic signal respectively to obtain a first single-point electrophysiological feature and a second single-point electrophysiological feature; Evaluate the muscle state according to the electromyographic signal to obtain a first evaluation result; evaluate the brain function state according to the electroencephalographic signal to obtain a second evaluation result; calculate the reaction time of the electromyographic signal and the electroencephalographic signal to obtain a third evaluation result; Inputting the acquired individual information, the first assessment result, the second assessment result, and the third assessment result into a fall risk assessment model, and outputting a risk coefficient; Inputting the multimodal electrophysiological feature, the first single-point electrophysiological feature, the second single-point electrophysiological feature, and the risk coefficient into a fall warning model for training to obtain a trained fall warning model; The multimodal electrophysiological signal of the target user is obtained, the multimodal electrophysiological signal of the target user is predicted based on the trained fall warning model, and a fall warning result is output.

[0007] In the second aspect, the present application proposes a multi-electrophysiological based fall warning system, comprising: Acquisition module: acquiring multimodal electrophysiological signals, wherein the multimodal electrophysiological signals include electromyographic signals and electroencephalographic signals; Feature extraction module: extracting features from the multimodal electrophysiological signal to obtain multimodal electrophysiological features; extracting features from the electromyographic signal and the electroencephalographic signal to obtain a first single-point electrophysiological feature and a second single-point electrophysiological feature; Evaluation module: evaluate the muscle state according to the electromyographic signal to obtain a first evaluation result; evaluate the brain function state according to the electroencephalographic signal to obtain a second evaluation result; calculate the reaction time of the electromyographic signal and the electroencephalographic signal to obtain a third evaluation result; input the acquired individual information, the first evaluation result, the second evaluation result, and the third evaluation result into a fall risk assessment model, and output a risk coefficient; Model training module: inputting the multimodal electrophysiological feature, the first single-point electrophysiological feature, the second single-point electrophysiological feature and the risk coefficient into a fall warning model for training to obtain a trained fall warning model; Output module: obtains the multimodal electrophysiological signal of the target user, predicts the multimodal electrophysiological signal of the target user based on the trained fall warning model, and outputs the fall warning result.

[0008] In a third aspect, the present application provides a fall warning device, which includes a processor and a memory coupled to the processor, wherein the memory stores program instructions for implementing a fall warning method based on multi-electrophysiology; and the processor is used to execute the program instructions stored in the memory to implement a fall warning method based on multi-electrophysiology.

[0009] In a fourth aspect, the present application provides a storage medium storing program instructions executable by a processor, wherein the program instructions are used to execute a multi-electrophysiological based fall warning method.

[0010] The present application provides a multi-electrophysiological fall warning method and system, which has the following beneficial effects: Through the multi-position electromyographic signals and 16-channel electroencephalographic signals caused by the human body's stress response, falls are warned, and the physiological and movement characteristics before falls are captured from multiple angles. A multimodal electrophysiological sensor network is used, and time synchronization is performed to obtain multimodal electrophysiological signals at the same time. Single-point electrophysiological feature extraction and multimodal electrophysiological feature extraction are used, which not only considers both irrelevant and relevant features, but also reduces the misjudgment rate of falls and improves the predictability of warnings. The trained fall warning model can accurately predict and can provide early warning and reaction time compared with traditional post-fall detection. At the same time, it can be adjusted according to individual differences and is suitable for different user groups. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic diagram of a flow chart of a multi-electrophysiological based fall warning method according to Embodiment 1 of the present application; Figure 2 This is a schematic diagram of the structure of a convolutional neural network for extracting features from multimodal electrophysiological signals according to Example 1 of the present application; Figure 3 This is a schematic diagram of the structure of a deep neural network for extracting features from electromyographic signals according to Example 1 of the present application; Figure 4 This is a schematic diagram of the structure of a convolutional neural network for extracting features from EEG signals according to Example 1 of the present application; Figure 5 This is a schematic diagram of the structure of the recurrent neural network of Example 1 of the present application; Figure 6 The target network model composed of a deep neural network, a convolutional neural network and a recurrent neural network in Example 1 of the present application; Figure 7 This is a structural schematic diagram of a multi-electrophysiological based fall warning system according to Example 2 of the present application; Figure 8 This is a design block diagram of a multi-electrophysiological based fall warning system according to Example 2 of the present application; Fig. 9 This is a structural diagram of a fall warning device according to Example 3 of the present application; Fig.10 This is a schematic diagram of the storage medium structure of Example 4 of the present application. DETAILED DESCRIPTION

[0012] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0013] The following is an analysis of the solutions in the prior art in combination with related technologies.

[0014] Invention patent CN118731940A discloses a method and system for intelligent fall detection based on millimeter wave radar, which involves the field of millimeter wave radar technology. It transmits linear frequency modulation wave signals to the target human body through millimeter wave radar, and receives electromagnetic wave signals reflected by the target human body. It uses artificial intelligence technology based on deep learning to analyze the time-frequency characteristics of linear frequency modulation wave signals and electromagnetic wave reflection signals, and then realizes intelligent recognition of the target human body posture based on the characteristic difference between the two. The detection method based on physical information such as millimeter waves and videos in this patent requires the help of environmental sensors, and sensor points need to be arranged in the environment in advance for detection. In addition, remote connection is required when linking with protective equipment, and real-time performance cannot be guaranteed.

[0015] Invention patent CN118609215A discloses a fall warning method and system based on deep learning, which belongs to the field of fall detection and image recognition technology. It includes entering the body parameters of the monitored target person, obtaining the actual center of mass coordinates based on the body parameters, using the first radar fall estimation model to perform indoor fall phenomenon detection, and using the second video stream iterative detection model to perform outdoor fall phenomenon, and continuously updating the training sample set to update and train the second video stream iterative detection model. However, this method can only be applied to environments where camera points are arranged in advance, and the real-time performance is not high, and it cannot be linked with protective equipment.

[0016] Invention patent CN118506807A discloses a method for detecting elderly falls based on deep learning and audio recognition, the method comprising: S1, constructing a fall audio data set; S2, performing data preprocessing to obtain a Mel-spectrogram with a unified format and containing audio time, frequency and volume feature information; S3, constructing a fall audio recognition model; S4, defining evaluation indicators for model training; S5, testing model accuracy. However, detection methods based on physical information such as audio are greatly affected by noise, and most of them are based on single information for detection, which is prone to misjudgment.

[0017] Therefore, this application proposes a fall warning method and system based on multi-electrophysiology. As we all know, when a person is about to fall, the body muscles will have an emergency reaction and the person will feel nervous. By accurately detecting the emergency reaction of the muscles through electromyographic signals and judging the nervousness through electroencephalographic signals, an early warning can be output in time to facilitate the protective equipment to take protective actions.

[0018] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0019] Example 1 See also Figure 1 , is a flow chart of a multi-electrophysiological fall warning method according to Example 1 of the present application; the steps include: S1: Acquire multimodal electrophysiological signals, where the multimodal electrophysiological signals include electromyographic signals and electroencephalographic signals.

[0020] In this embodiment, the locations for collecting the electromyographic signals include but are not limited to: lower limb muscles, upper limb muscles, buttocks muscles, core muscles, back muscles, and neck muscles; the electroencephalographic signal collection uses 16 channels. The electromyographic collection at different locations and the 16-channel electroencephalographic collection are networked using Bluetooth to obtain a multimodal electrophysiological human sensor network. The various electrophysiological collection points between the multimodal electrophysiological human sensor network have a time synchronization function, and multimodal electrophysiological signals at the same time can be obtained. The specific implementation method is as follows: First, Bluetooth networking is performed for electromyography and electroencephalography acquisition, and the respective locations and acquisition types are determined. Each Bluetooth module has its own MAC address, which is bound to the acquisition location and type of the respective electrophysiological signal. A node is added or deleted according to the different situations of the user, and the information between nodes can be communicated after networking. The implementation example is shown in the following table:

[0021] To synchronize timestamps between nodes, first select a reference node, whose timestamp is the reference timestamp Tref, and then synchronize the timestamps of the nodes according to the reference timestamp. , get the time offset of each node ( is the number of nodes). In the subsequent data collection, the data of each node is time-shifted accordingly. The details are as follows: , , in For the Data after synchronization of nodes; For the The original data of each node; It is a synchronization function, which is accomplished by shifting the sampling points.

[0022] The synchronized data is preprocessed, mainly by using a bandpass filter to filter out the frequency range of interest in the electrophysiological signal, the EMG range is 20~500Hz, and the EEG range is between 0.5~100Hz. Then a notch filter is used to filter out the power frequency noise (50Hz or 60Hz).

[0023] S2: extracting features from the multimodal electrophysiological signal to obtain a multimodal electrophysiological feature; extracting features from the electromyographic signal and the electroencephalographic signal to obtain a first single-point electrophysiological feature and a second single-point electrophysiological feature.

[0024] In this embodiment, feature extraction mainly includes two parts: single-point electrophysiological feature extraction and multimodal electrophysiological feature extraction. Single-point electrophysiological feature extraction can accurately extract different single-point electrophysiological features without considering other electrophysiological features, and multimodal electrophysiological feature extraction can extract correlation features between electromyographic signals at different positions and 16-channel EEG signals.

[0025] See also Figure 2 , which is a schematic diagram of the structure of the convolutional neural network for feature extraction of multimodal electrophysiological signals in Example 1 of the present application. Since the multimodal electrophysiological signal data is large, the network design is more complicated. It mainly includes a multi-position electromyography convolution pooling layer with a convolution kernel of 4x4, a multi-channel electroencephalogram convolution pooling layer with a convolution kernel of 4x4, a comprehensive convolution pooling layer with a convolution kernel of 3x3, and three fully connected layers with 64, 256, and 9 nodes respectively. The output is the coordination degree of muscles and brain in 8 positions and the current stability.

[0026] The convolutional neural network is pre-trained through electromyographic signals and multi-channel electroencephalographic signals at various positions, and multimodal electrophysiological features are output, that is, the coordination degree of muscles and brain at various positions and the current degree of stability are obtained, that is, multimodal electrophysiological features are obtained.

[0027] See also Figure 3 , which is a schematic diagram of the structure of the deep neural network for feature extraction of electromyographic signals in Example 1 of the present application. Since the complexity of electromyographic signals is not very high, the network structure does not need to be too complex. A deep neural network is used to extract features from electromyographic signals. The deep neural network is composed entirely of fully connected layers (FCLs), which are divided into 4 layers. The number of nodes in each layer is 64, 256, 128, and 3, respectively. The output is the degree of muscle contraction, fatigue, and lactic acid accumulation, that is, the first single-point electrophysiological feature is obtained.

[0028] See also Figure 4, which is a schematic diagram of the structure of the convolutional neural network for feature extraction of EEG signals in Example 1 of the present application. Since EEG data is multi-channel data, a convolutional neural network is used for feature extraction. The convolutional neural network is divided into four layers. The first two layers are convolution pooling layers (conv pool), which mainly reduce the dimension of the 16-channel EEG signals twice. The first convolution pooling layer uses a convolution kernel of 4x4, and the second convolution pooling layer uses a convolution kernel of 3x3; the last two layers are fully connected layers (FCL), with 128 and 3 nodes respectively, and the outputs are the degree of emotional tension, the degree of emotional relief, and the current degree of attention. That is, the second single-point electrophysiological feature is obtained.

[0029] S3: Evaluate the muscle state according to the electromyographic signal to obtain a first evaluation result; evaluate the brain function state according to the electroencephalographic signal to obtain a second evaluation result; calculate the reaction time of the electromyographic signal and the electroencephalographic signal to obtain a third evaluation result.

[0030] In this embodiment, since the muscle state, brain function state, fall risk factor and type will not change suddenly, a periodic assessment is performed, that is, an assessment is performed after obtaining the electromyographic signals and electroencephalographic signals for a week, and the assessment results are stored.

[0031] The electromyographic signal is input into a support vector machine to evaluate the muscle state, and a first evaluation result is output, which includes a calculation method based on root mean square, integrated electromyographic value and median frequency; as muscle function weakens, the root mean square RMS will gradually decrease, the integrated electromyographic value IEMG will also gradually decrease, and the median frequency MF will gradually decrease.

[0032] The calculation formula of the root mean square is: , in For each point of the EMG signal, is the number of sampling points of the electromyographic signal; The calculation formula of the integrated electromyographic value is: , When calculating the median frequency, you need to use FFT to calculate the power spectral density: , in is the power spectral density, is the frequency, is the time series of electromyographic signal, and FFT is fast Fourier transform; The median frequency is calculated by the power spectral density, and the calculation formula is: .

[0033] Furthermore, the brain function state is evaluated by random forest, and a second evaluation result is output. Specifically: Obtain the intensity of different waves based on the power spectral density, including Wave, Wave, Wave, Wave. The waves are related to basal neural activity and postural stability; waves are associated with attention and postural control; waves reflect brain inhibition and sensor integration; Waves are associated with active movement and postural regulation.

[0034] The power spectral density of the wave is expressed as: , The power spectral density of the wave is expressed as: , The power spectral density of the wave is expressed as: , The power spectral density of the wave is expressed as: , in is the time series of EEG signals, and FFT is fast Fourier transform.

[0035] At the same time, the phase-locked value PLV is used to calculate the phase synchronization between brain regions and evaluate the functional connectivity of brain regions: , in is the number of sampling points of the EEG signal, is the instantaneous phase of the first EEG signal at the nth time point, is the instantaneous phase of the second signal at the nth time point, is a natural constant, Is an imaginary unit.

[0036] Furthermore, in addition to the RMS, IEMG, and MF of the electromyographic signal and the four-wave power spectrum and PLV of the EEG signal, it is also necessary to calculate the electromyographic reaction time. and EEG reaction time , giving specific stimulation during the evaluation. The calculation formula is: , , in is the muscle reaction time; Provide muscle stimulation time; is the EEG response time, i.e., the peak time of the P300 wave; Stimulation time for brain areas.

[0037] S4: Input the acquired individual information, the first assessment result, the second assessment result, and the third assessment result into a fall risk assessment model, and output a risk coefficient.

[0038] In this embodiment, the first assessment result, the second assessment result, the third assessment result and the acquired individual information are input into a fall risk assessment model for fall risk assessment, and the fall risk assessment model is a deep neural network. The deep neural network consists of 4 fully connected layers (FCL), with the number of nodes being 128, 256, 64, and 4, respectively, and the output is the risk coefficient of forward, backward, left, and right falls.

[0039] S5: Inputting the multimodal electrophysiological feature, the first single-point electrophysiological feature, the second single-point electrophysiological feature, and the risk coefficient into a fall warning model for training to obtain a trained fall warning model.

[0040] In this embodiment, the fall warning model is a recurrent neural network, wherein the risk factor, the convolutional neural network for extracting features from the multimodal electrophysiological signals, the deep neural network for extracting features from the electromyographic signals, and the convolutional neural network for extracting features from the electroencephalographic signals are locked and then connected to the fall warning model; after the connection, the fall warning model is trained to obtain a trained fall warning model.

[0041] See also Figure 5 , which is a schematic diagram of the structure of the recurrent neural network of Example 1 of the present application. The recurrent neural network includes: a fully connected layer with 128 nodes, three LSTM layers with 64, 64, and 32 nodes respectively, and a fully connected layer with 1 output node.

[0042] See also Figure 6 , is the target network model composed of a deep neural network, a convolutional neural network and a recurrent neural network in Example 1 of the present application. The target network model is composed of DNN, CNN and RNN, and continuously collects data from the normal state to the falling process to retrain the target network model. At this time, there is no need to lock any parameters, and the target network model is directly trained in real time. The weights and biases are further optimized during training. The specific formula is:

[0043]

[0044] in is the optimized weight, is the pre-trained weight, Optimize the value for the weight, is the optimized bias, is the pre-training bias, is the optimized value for bias.

[0045] S6: Acquire the multimodal electrophysiological signal of the target user, predict the multimodal electrophysiological signal of the target user based on the trained fall warning model, and output a fall warning result.

[0046] In this embodiment, the trained fall warning model can be adjusted according to individual differences of users and is suitable for different user groups.

[0047] In summary, the present application uses multi-position electromyographic signals and 16-channel electroencephalographic signals caused by the human body's stress response to warn of falls, capturing the physiological and movement characteristics before a fall from multiple angles; a multimodal electrophysiological sensing network is adopted, and time synchronization is performed to obtain multimodal electrophysiological signals at the same time; single-point electrophysiological feature extraction and multimodal electrophysiological feature extraction are adopted at the same time, which takes into account both irrelevant and relevant features, reduces the misjudgment rate of falls, and improves the predictability of warnings; signal acquisition increases the real-time nature of warnings for human-body signal acquisition, can provide early warnings, and provide reaction time; and the fall warning model can be self-trained and adjusted according to individual differences, and can be suitable for different user groups.

[0048] Example 2 See also Figure 7 , is a structural diagram of a multi-electrophysiological fall warning system according to Example 2 of the present application; the specific contents include: Acquisition module: acquiring multimodal electrophysiological signals, wherein the multimodal electrophysiological signals include electromyographic signals and electroencephalographic signals; Feature extraction module: extracting features from the multimodal electrophysiological signal to obtain multimodal electrophysiological features; extracting features from the electromyographic signal and the electroencephalographic signal to obtain a first single-point electrophysiological feature and a second single-point electrophysiological feature; Evaluation module: evaluate the muscle state according to the electromyographic signal to obtain a first evaluation result; evaluate the brain function state according to the electroencephalographic signal to obtain a second evaluation result; calculate the reaction time of the electromyographic signal and the electroencephalographic signal to obtain a third evaluation result; input the acquired individual information, the first evaluation result, the second evaluation result, and the third evaluation result into a fall risk assessment model, and output a risk coefficient; Model training module: inputting the multimodal electrophysiological feature, the first single-point electrophysiological feature, the second single-point electrophysiological feature and the risk coefficient into a fall warning model for training to obtain a trained fall warning model; Output module: obtains the multimodal electrophysiological signal of the target user, predicts the multimodal electrophysiological signal of the target user based on the trained fall warning model, and outputs the fall warning result.

[0049] In this embodiment, after the training module is completed, the system enters the use phase and issues an early warning for falls. The system can be linked with protective equipment.

[0050] See also Figure 8 , which is a design block diagram of a fall warning system based on multi-electrophysiology in Example 2 of the present application. It mainly includes two stages. The first stage is the individual assessment stage. The muscle state assessment is used to assess the health of the muscles, the single-point brain function assessment is used to assess the user's brain function state, and the fall risk assessment based on multimodal electrophysiology and individual conditions, and the risk coefficient and type are saved. The second stage is the warning stage, including a multimodal sensing network composed of multi-position electromyography and multi-channel EEG, single-point electrophysiological feature extraction, multimodal electrophysiological feature extraction, and the muscle state, brain function state, and fall risk coefficient and type saved in the first stage are input into the fall warning model for training, so that the trained model can provide real-time warning.

[0051] Example 3 See also Fig. 9 , is a schematic diagram of the structure of a fall warning device according to Embodiment 3 of the present application. The fall warning device 50 includes a processor 51 and a memory 52 coupled to the processor 51 .

[0052] The memory 52 stores program instructions for implementing the above-mentioned multi-electrophysiological based fall warning method.

[0053] The processor 51 is used to execute program instructions stored in the memory 52 to implement a multi-electrophysiological based fall warning.

[0054] The processor 51 may also be referred to as a CPU (Central Processing Unit).

[0055] The processor 51 may be an integrated circuit chip with signal processing capability. The processor 51 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0056] Example 4 See also Fig.10, which is a schematic diagram of the structure of the storage medium of Example 4 of the present application. The storage medium of the embodiment of the present application stores a program file 61 that can implement all the above methods, wherein the program file 61 can be stored in the above storage medium in the form of a software product, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, or a computer, server, mobile phone, tablet and other devices.

[0057] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0058] The above description is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

[0059] Although the embodiments of the present application have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the appended claims and their equivalents.

[0060] Of course, the present invention may have many other implementations. Based on this implementation, other implementations obtained by ordinary technicians in this field without any creative work are all within the scope of protection of the present invention.

Claims

1. A fall warning method based on multi-electrophysiology, characterized in that: include: Acquiring multimodal electrophysiological signals, wherein the multimodal electrophysiological signals include electromyographic signals and electroencephalographic signals; Extracting features from the multimodal electrophysiological signals to obtain multimodal electrophysiological features; Perform feature extraction on the electromyographic signal and the electroencephalographic signal respectively to obtain a first single-point electrophysiological feature and a second single-point electrophysiological feature; Evaluate the muscle state according to the electromyographic signal to obtain a first evaluation result; Evaluate the brain function state according to the electroencephalogram signal to obtain a second evaluation result; Calculating the reaction time of the electromyographic signal and the electroencephalographic signal to obtain a third evaluation result; Inputting the acquired individual information, the first assessment result, the second assessment result, and the third assessment result into a fall risk assessment model, and outputting a risk coefficient; Inputting the multimodal electrophysiological feature, the first single-point electrophysiological feature, the second single-point electrophysiological feature, and the risk coefficient into a fall warning model for training to obtain a trained fall warning model; The multimodal electrophysiological signal of the target user is obtained, the multimodal electrophysiological signal of the target user is predicted based on the trained fall warning model, and a fall warning result is output.

2. A multi-electrophysiological fall warning method according to claim 1, characterized in that: The step of acquiring a multimodal electrophysiological signal, wherein the multimodal electrophysiological signal includes an electromyographic signal and an electroencephalographic signal, specifically includes the following steps: The locations where the electromyographic signals are collected include but are not limited to: lower limb muscles, upper limb muscles, hip muscles, core muscles, back muscles, and neck muscles; the electroencephalographic signals are collected using 16 channels; The myoelectric acquisition at different positions is networked with the 16-channel EEG acquisition to obtain a multimodal electrophysiological human body sensor network. The electrophysiological acquisition points between the multimodal electrophysiological human body sensor network are synchronized with timestamps, and the data after the timestamp synchronization are preprocessed to obtain multimodal electrophysiological signals at the same time, that is, electromyographic signals and electroencephalographic signals.

3. A multi-electrophysiological fall warning method according to claim 1, characterized in that: The step of extracting features from the multimodal electrophysiological signals to obtain multimodal electrophysiological features specifically includes the following steps: A convolutional neural network is used to extract features from the multimodal electrophysiological signal; wherein the convolutional neural network includes but is not limited to: a multi-position electromyography convolutional pooling layer, a multi-channel electroencephalography convolutional pooling layer, a comprehensive convolutional pooling layer, and a fully connected layer; The convolutional neural network is pre-trained through electromyographic signals and multi-channel electroencephalographic signals at various positions, and multimodal electrophysiological characteristics are output, that is, the coordination degree of muscles and brain at various positions and the current stability are obtained.

4. The multi-electrophysiological fall warning method according to claim 1, characterized in that: The step of extracting features from the electromyographic signal and the electroencephalographic signal to obtain a first single-point electrophysiological feature and a second single-point electrophysiological feature specifically includes the following steps: A deep neural network is used to extract features from the electromyographic signal, wherein the deep neural network is composed of four fully connected layers; the electromyographic signal is input into the deep neural network for pre-training, and a first single-point electrophysiological feature is output, that is, the degree of muscle contraction, the degree of fatigue, and the degree of lactic acid accumulation are obtained; A convolutional neural network is used to extract features from the EEG signal, wherein the convolutional neural network consists of two convolutional pooling layers and two fully connected layers, and the two convolutional pooling layers perform two dimensionality reduction on the EEG signal; the EEG signal is input into the convolutional neural network for pre-training, and a second single-point electrophysiological feature is output, namely, the degree of emotional tension, the degree of emotional relief, and the current degree of attention are obtained.

5. The multi-electrophysiological fall warning method according to claim 1, characterized in that: The step of evaluating the muscle state according to the electromyographic signal to obtain a first evaluation result specifically includes the following steps: Inputting the electromyographic signal into a support vector machine to evaluate the muscle state, and outputting a first evaluation result; including a calculation method based on root mean square, integrated electromyographic value and median frequency; The calculation formula of the root mean square is: , in For each point of the EMG signal, is the number of sampling points of the electromyographic signal; The calculation formula of the integrated electromyographic value is: , Use FFT to calculate the power spectral density: , in is the power spectral density, is the frequency, is the time series of electromyographic signal, and FFT is fast Fourier transform; The median frequency is calculated by the power spectral density, and the calculation formula is: .

6. The multi-electrophysiological fall warning method according to claim 1, characterized in that: The step of evaluating the brain function state according to the electroencephalogram signal to obtain a second evaluation result specifically includes the following steps: The brain function status is evaluated by random forest, and the second evaluation result is output; specifically: Obtain the intensity of different waves based on the power spectral density, including Wave, Wave, Wave, Wave; Said The power spectral density of the wave is expressed as: , Said The power spectral density of the wave is expressed as: , Said The power spectral density of the wave is expressed as: , Said The power spectral density of the wave is expressed as: , in is the time series of EEG signals, and FFT is fast Fourier transform; Phase-locked value PLV was used to calculate the phase synchronization between brain regions and evaluate the functional connectivity of brain regions: ， in is the number of sampling points of the EEG signal, is the instantaneous phase of the first EEG signal at the nth time point, is the instantaneous phase of the second signal at the nth time point, is a natural constant, Is an imaginary unit.

7. The multi-electrophysiological fall warning method according to claim 1, characterized in that: The fall risk assessment model is a deep neural network, which inputs the acquired individual information, the first assessment result, the second assessment result, and the third assessment result into the fall risk assessment model, and outputs the risk coefficients of forward, backward, left, and right falls; The fall warning model is a recurrent neural network, wherein the risk factor, the convolutional neural network for extracting features from the multimodal electrophysiological signals, the deep neural network for extracting features from the electromyographic signals, and the convolutional neural network for extracting features from the electroencephalographic signals are connected to the fall warning model after being locked; after the connection, the fall warning model is trained to obtain a trained fall warning model.

8. A system for a fall warning method based on multi-electrophysiology according to claim 1, characterized in that: include: Acquisition module: acquiring multimodal electrophysiological signals, wherein the multimodal electrophysiological signals include electromyographic signals and electroencephalographic signals; Feature extraction module: extracting features from the multimodal electrophysiological signals to obtain multimodal electrophysiological features; Perform feature extraction on the electromyographic signal and the electroencephalographic signal respectively to obtain a first single-point electrophysiological feature and a second single-point electrophysiological feature; Evaluation module: evaluates the muscle state according to the electromyographic signal to obtain a first evaluation result; Evaluate the brain function state according to the electroencephalogram signal to obtain a second evaluation result; Calculating the reaction time of the electromyographic signal and the electroencephalographic signal to obtain a third assessment result; inputting the acquired individual information, the first assessment result, the second assessment result, and the third assessment result into a fall risk assessment model to output a risk coefficient; Model training module: inputting the multimodal electrophysiological feature, the first single-point electrophysiological feature, the second single-point electrophysiological feature and the risk coefficient into a fall warning model for training to obtain a trained fall warning model; Output module: obtains the multimodal electrophysiological signal of the target user, predicts the multimodal electrophysiological signal of the target user based on the trained fall warning model, and outputs the fall warning result.

9. A fall warning device, characterized in that: The fall warning device includes a processor and a memory coupled to the processor, wherein the memory stores program instructions for implementing a multi-electrophysiological based fall warning method as described in any one of claims 1-7; and the processor is used to execute the program instructions stored in the memory to implement a multi-electrophysiological based fall warning.

10. A storage medium, characterized in that: Program instructions executable by a processor are stored, and the program instructions are used to execute a multi-electrophysiological based fall warning method as described in any one of claims 1-7.

Images