User classification method, device and electronic device for blood pressure measuring device

Abstract

The present application discloses a user classification method and device for a blood pressure measuring device, which belongs to the field of computer technology. The method comprises: obtaining a pressure signal and a piezoelectric signal collected by the blood pressure measuring device during the blood pressure measurement process of the current user; performing signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation; performing feature extraction on the sequence through a pre-trained feature extraction network to obtain the current blood pressure feature; and determining whether the current user is a designated registered user or a guest user based on the current blood pressure feature and the registered blood pressure feature of the registered user of the blood pressure measuring device. The present method realizes the distinction between different measured persons based on blood pressure features, thereby achieving the non-interference of the blood pressure data of the measured persons, and the unlimited number of registered users. The blood pressure measuring device using the present method does not need to manually switch users when performing blood pressure measurement, and the operation is more convenient.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and apparatus for classifying users of a blood pressure measuring device, an electronic device, and a computer-readable storage medium. Background Art

[0002] A blood pressure monitor can accurately record a person's blood pressure changes over time, providing valuable insights for both the person being measured and their physician. However, in practice, a household often only has one blood pressure monitor. Therefore, user classification plays a crucial role in analyzing a person's long-term blood pressure. Prior art blood pressure monitors use a single DIP switch to classify users as either A or B. However, this user classification method has at least the following drawbacks: 1. Using a DIP switch to distinguish between users A and B limits the number of users that can be classified; 2. It is susceptible to interference from blood pressure data from other users; and 3. The user interface is cumbersome.

[0003] It can be seen that the user classification method of the blood pressure measurement device in the prior art needs to be improved. Summary of the Invention

[0004] The embodiments of the present application provide a method and apparatus for classifying users of a blood pressure measuring device, which is used to solve the problems that the number of user categories of a blood pressure measuring device is limited, the measurement data is easily interfered with by the measurement data of other users, and the device is cumbersome to use.

[0005] In a first aspect, an embodiment of the present application provides a method for classifying users of a blood pressure measurement device, comprising:

[0006] Obtaining a pressure signal and a piezoelectric signal collected by a blood pressure measuring device during a blood pressure measurement process for a current user;

[0007] performing signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation;

[0008] Extract features from the sequence using a pre-trained feature extraction network to obtain current blood pressure features;

[0009] The user category that matches the current user is determined based on the current blood pressure characteristics and the registered blood pressure characteristics of the registered user of the blood pressure measurement device, wherein the user category includes: specifying the registered user or the guest user.

[0010] Optionally, determining a user category that matches the current user based on the current blood pressure characteristics and the registered blood pressure characteristics of a registered user of the blood pressure measurement device includes:

[0011] Obtaining a similarity distance between a registered blood pressure feature of a registered user of the blood pressure measurement device and the current blood pressure feature;

[0012] Determine the registered user to whom the registered blood pressure feature corresponding to the minimum similarity distance belongs as a candidate registered user;

[0013] The user category that the current user matches is determined based on the magnitude relationship between the similarity distance threshold corresponding to the candidate registered user and the similarity distance minimum value.

[0014] Optionally, determining the user category that the current user matches based on the relationship between the similarity distance threshold corresponding to the candidate registered user and the minimum similarity distance value includes:

[0015] Adjusting the similarity distance threshold according to a preset ratio value to obtain a reference threshold, wherein the preset ratio value is greater than or equal to 1;

[0016] In response to the minimum similarity distance being less than or equal to the reference threshold, determining that the user category matched by the current user is the candidate registered user;

[0017] In response to the minimum similarity distance being greater than the reference threshold, it is determined that the current user matches the guest user.

[0018] Optionally, the registered blood pressure characteristics of the registered user are pre-determined by the following method:

[0019] In the user registration mode, obtaining pressure signals and piezoelectric signals collected by the blood pressure measurement device during N blood pressure measurement processes of the registered user, and obtaining blood pressure characteristics corresponding to each blood pressure measurement process based on the pressure signals and the piezoelectric signals;

[0020] The registered blood pressure feature of the registered user is obtained according to the average feature of the blood pressure features corresponding to N blood pressure measurement processes, where N is an integer greater than 1.

[0021] Optionally, the similarity distance threshold corresponding to the candidate registered user is pre-determined by the following method:

[0022] A similarity distance threshold corresponding to the candidate registered user is obtained according to the maximum similarity distance between the blood pressure feature corresponding to the N blood pressure measurement processes of the candidate registered user and the average feature.

[0023] Optionally, the pressure signal is a pressure signal within an airbag of a cuff of the blood pressure measuring device, and the piezoelectric signal is collected by a piezoelectric device disposed on a side of the cuff that contacts the brachial artery of the current user. The signal characteristics include: a pressure wave difference, a first piezoelectric signal characteristic, a second piezoelectric signal characteristic, and an acquisition time characteristic corresponding to a brachial artery pulsation. Signal processing is performed on the pressure signal and the piezoelectric signal to obtain a sequence of signal characteristics corresponding to each brachial artery pulsation, including:

[0024] Band-pass filtering is performed on the pressure signal to obtain an oscillation wave signal within a preset frequency range;

[0025] Determining the acquisition time of the pressure signal corresponding to each peak value of the shock wave signal;

[0026] Determining, based on the acquisition time, the acquisition time feature corresponding to the corresponding brachial artery pulsation;

[0027] determining a pressure wave difference corresponding to the acquisition time according to a change amplitude of a pressure signal in the shock wave signal corresponding to the acquisition time;

[0028] determining a first piezoelectric signal feature corresponding to the acquisition time according to a first frequency band piezoelectric signal obtained by performing bandpass filtering on the piezoelectric signal;

[0029] determining a second piezoelectric signal feature corresponding to the acquisition time based on a second frequency band piezoelectric signal obtained by performing segmented bandpass filtering on the piezoelectric signal, wherein the frequency of the first frequency band piezoelectric signal is lower than that of the second frequency band piezoelectric signal;

[0030] The acquisition time feature, the pressure wave difference, the first piezoelectric signal feature, and the second piezoelectric signal feature corresponding to each acquisition time are used in the order of the acquisition time to generate a sequence of the signal features.

[0031] Optionally, the feature extraction network includes: a bidirectional long short-term memory network and a feature conversion layer. The feature extraction of the sequence by the pre-trained feature extraction network to obtain the current blood pressure feature includes:

[0032] Inputting the sequence into the bidirectional long short-term memory network, performing feature encoding on each signal feature in the sequence through the bidirectional long short-term memory network, and obtaining a hidden layer vector corresponding to each signal feature;

[0033] The feature conversion layer is called to perform an arithmetic mean operation on the hidden layer vector of the signal feature to obtain the current blood pressure feature.

[0034] Optionally, the feature extraction network is obtained by training a user identification model based on a blood pressure signal, wherein the user identification model based on a blood pressure signal comprises: the feature extraction network and a fully connected layer, and the user identification model based on a blood pressure signal is trained by the following method:

[0035] Acquire a number of training samples, wherein the sample data of each training sample is: a sequence of signal features obtained by processing the pressure signal and piezoelectric signal collected based on a single blood pressure measurement, and the sample label is: the true value of the user category;

[0036] For each of the training samples, extracting features from the sample data using the feature extraction network, and using the hidden layer vector obtained by the feature extraction as the blood pressure feature corresponding to the training sample;

[0037] Classify and map the blood pressure features through the fully connected layer to obtain a user category prediction value corresponding to the training sample;

[0038] Calculating a model loss of the user recognition model based on the blood pressure signal according to the user category prediction value and the sample label;

[0039] The user recognition model based on the blood pressure signal is iteratively trained by optimizing the model loss to obtain the feature extraction network.

[0040] In a second aspect, an embodiment of the present application provides a user classification device for a blood pressure measurement device, comprising:

[0041] A blood pressure measurement signal acquisition module is used to obtain a pressure signal and a piezoelectric signal collected by the blood pressure measurement device during the blood pressure measurement process of the current user;

[0042] a signal feature sequence acquisition module, configured to perform signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation;

[0043] A current blood pressure feature acquisition module is used to extract features from the sequence using a pre-trained feature extraction network to obtain current blood pressure features;

[0044] The user classification module is configured to determine a user category that matches the current user based on the current blood pressure characteristics and the registered blood pressure characteristics of the registered user of the blood pressure measuring device, wherein the user category includes: specifying the registered user or the guest user.

[0045] Optionally, the user identification module is further configured to:

[0046] Obtaining a similarity distance between a registered blood pressure feature of a registered user of the blood pressure measurement device and the current blood pressure feature;

[0047] Determine the registered user to whom the registered blood pressure feature corresponding to the minimum similarity distance belongs as a candidate registered user;

[0048] The user category that the current user matches is determined based on the magnitude relationship between the similarity distance threshold corresponding to the candidate registered user and the similarity distance minimum value.

[0049] Optionally, determining the user category that the current user matches based on the relationship between the similarity distance threshold corresponding to the candidate registered user and the minimum similarity distance value includes:

[0050] Adjusting the similarity distance threshold according to a preset ratio value to obtain a reference threshold, wherein the preset ratio value is greater than or equal to 1;

[0051] In response to the minimum similarity distance being less than or equal to the reference threshold, determining that the user category matched by the current user is the candidate registered user;

[0052] In response to the minimum similarity distance being greater than the reference threshold, it is determined that the current user matches the guest user.

[0053] Optionally, the registered blood pressure characteristics of the registered user are pre-determined by the following method:

[0054] In the user registration mode, obtaining pressure signals and piezoelectric signals collected by the blood pressure measurement device during N blood pressure measurement processes of the registered user, and obtaining blood pressure characteristics corresponding to each blood pressure measurement process based on the pressure signals and the piezoelectric signals;

[0055] The registered blood pressure feature of the registered user is obtained according to the average feature of the blood pressure features corresponding to N blood pressure measurement processes, where N is an integer greater than 1.

[0056] Optionally, the similarity distance threshold corresponding to the candidate registered user is pre-determined by the following method:

[0057] A similarity distance threshold corresponding to the candidate registered user is obtained according to the maximum similarity distance between the blood pressure feature corresponding to the N blood pressure measurement processes of the candidate registered user and the average feature.

[0058] Optionally, the pressure signal is a pressure signal within an airbag of a cuff of the blood pressure measuring device, the piezoelectric signal is collected by a piezoelectric device disposed on a side of the cuff contacting the brachial artery of the current user, and the signal characteristics include: a pressure wave difference, a first piezoelectric signal characteristic, a second piezoelectric signal characteristic, and an acquisition time characteristic corresponding to the brachial artery pulsation. The signal characteristic sequence acquisition module is further configured to:

[0059] Band-pass filtering is performed on the pressure signal to obtain an oscillation wave signal within a preset frequency range;

[0060] Determining the acquisition time of the pressure signal corresponding to each peak value of the shock wave signal;

[0061] Determining, based on the acquisition time, the acquisition time feature corresponding to the corresponding brachial artery pulsation;

[0062] determining a pressure wave difference corresponding to the acquisition time according to a change amplitude of a pressure signal in the shock wave signal corresponding to the acquisition time;

[0063] determining a first piezoelectric signal feature corresponding to the acquisition time according to a first frequency band piezoelectric signal obtained by performing bandpass filtering on the piezoelectric signal;

[0064] determining a second piezoelectric signal feature corresponding to the acquisition time based on a second frequency band piezoelectric signal obtained by performing segmented bandpass filtering on the piezoelectric signal, wherein the frequency of the first frequency band piezoelectric signal is lower than that of the second frequency band piezoelectric signal;

[0065] The acquisition time feature, the pressure wave difference, the first piezoelectric signal feature, and the second piezoelectric signal feature corresponding to each acquisition time are used in the order of the acquisition time to generate a sequence of the signal features.

[0066] Optionally, the feature extraction network includes: a bidirectional long short-term memory network, a feature conversion layer, and the current blood pressure feature acquisition module is further used to:

[0067] Inputting the sequence into the bidirectional long short-term memory network, performing feature encoding on each signal feature in the sequence through the bidirectional long short-term memory network, and obtaining a hidden layer vector corresponding to each signal feature;

[0068] The feature conversion layer is called to perform an arithmetic mean operation on the hidden layer vector of the signal feature to obtain the current blood pressure feature.

[0069] Optionally, the feature extraction network is obtained by training a user identification model based on a blood pressure signal, wherein the user identification model based on a blood pressure signal comprises: the feature extraction network and a fully connected layer, and the user identification model based on a blood pressure signal is trained by the following method:

[0070] Acquire a number of training samples, wherein the sample data of each training sample is: a sequence of signal features obtained by processing the pressure signal and piezoelectric signal collected based on a single blood pressure measurement, and the sample label is: the true value of the user category;

[0071] For each of the training samples, extracting features from the sample data using the feature extraction network, and using the hidden layer vector obtained by the feature extraction as the blood pressure feature corresponding to the training sample;

[0072] Classify and map the blood pressure features through the fully connected layer to obtain a user category prediction value corresponding to the training sample;

[0073] Calculating a model loss of the user recognition model based on the blood pressure signal according to the user category prediction value and the sample label;

[0074] The user recognition model based on the blood pressure signal is iteratively trained by optimizing the model loss to obtain the feature extraction network.

[0075] In a third aspect, an embodiment of the present application further discloses an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the user classification method for the blood pressure measurement device described in the embodiment of the present application is implemented.

[0076] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the user classification method for a blood pressure measurement device disclosed in an embodiment of the present application.

[0077] The user classification method for a blood pressure measuring device disclosed in an embodiment of the present application obtains a pressure signal and a piezoelectric signal collected during the blood pressure measurement process of the current user by the blood pressure measuring device; performs signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation; extracts features from the sequence through a pre-trained feature extraction network to obtain current blood pressure features; determines a user category matching the current user based on the current blood pressure features and the registered blood pressure features of the registered user of the blood pressure measuring device, wherein the user category includes: specifying the registered user or the guest user, thereby achieving the distinction between different measured persons based on blood pressure features, solving the problems that the number of user categories of the blood pressure measuring device is limited, the measurement data is easily interfered with by the measurement data of other users, and the use is cumbersome.

[0078] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0080] Figure 1 This is a flow chart of a method for classifying users of a blood pressure measurement device according to an embodiment of the present application;

[0081] Figure 2 is a structural diagram of a blood pressure measurement device disclosed in an embodiment of the present application;

[0082] Figure 3 is another structural schematic diagram of the blood pressure measurement device disclosed in the embodiments of the present application;

[0083] Figure 4 This is a schematic diagram of the pressure signal waveform collected in the embodiment of the present application;

[0084] Figure 5 Schematic diagram of the piezoelectric signal waveform collected in the embodiment of the present application;

[0085] Figure 6 Schematic diagram of shock wave and bandpass pressure wave obtained after filtering in the embodiment of the present application;

[0086] Figure 7 This is a schematic diagram of the working principle of the user identification model based on blood pressure signals in an embodiment of the present application;

[0087] Figure 8 This is another flow chart of the method for classifying users of the blood pressure measurement device according to an embodiment of the present application;

[0088] Figure 9 This is a schematic diagram of the structure of a user classification device of a blood pressure measurement device in one embodiment of the present application;

[0089] Figure 10 A block diagram schematically shows an electronic device for executing the method according to the present application; and

[0090] Figure 11 The figure schematically shows a storage unit for storing or carrying a program code for implementing the method according to the present application. DETAILED DESCRIPTION

[0091] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0092] The embodiment of the present application discloses a method for classifying users of a blood pressure measuring device, such as Figure 1 As shown, the method includes: steps 110 to 140.

[0093] Step 110: Obtain a pressure signal and a piezoelectric signal collected by the blood pressure measuring device during the blood pressure measurement process of the current user.

[0094] First, the pressure signal and piezoelectric signal generated by the brachial artery pulsation collected by the cuff worn on the user's arm are obtained, wherein the pressure signal is the pressure signal in the air bag of the cuff, and the piezoelectric signal is collected by a piezoelectric device arranged on the side of the cuff that is attached to the user's arm.

[0095] The user classification method of the blood pressure measuring device disclosed in the embodiment of the present application is applied to a blood pressure measuring device including a cuff with an inflatable air bag, and the cuff needs to be provided with a pressure detection device and a piezoelectric device. Figure 2 and Figure 3 The structural diagram of the blood pressure measurement device shown in the figure illustrates a specific implementation method for obtaining the pressure signal and piezoelectric signal generated by the brachial artery pulsation collected by the cuff worn on the user's arm.

[0096] like Figure 2 As shown, the blood pressure measurement device includes: a cuff 200 and a signal output device 210, wherein the cuff 200 further includes: an air bag 201, a pressure detection device 202, and a piezoelectric device 203. The pressure detection device 202 and the piezoelectric device 203 are respectively connected to the signal output device 210 for communication.

[0097] Wherein, the pressure detection device 202 is configured to collect the pressure signal in the airbag. The pressure detection device 202 is arranged in the airbag 201. In some embodiments of the present application, the pressure detection device 202 can be implemented based on a pressure sensor device. In the process of measuring blood pressure, first, the airbag 201 is inflated and the cuff 200 is pressurized to block the blood flow in the upper arm artery. After that, the gas in the airbag 201 is slowly released at a rate of 2-4 mmHg / s. In the process of the airbag 201 releasing gas, the pressure detection device 202 will detect the corresponding pressure signal generated by the gas in the airbag due to the pulsation of the brachial artery.

[0098] The piezoelectric device 203 is arranged on the side of the cuff that fits the user's arm, and is configured to collect the piezoelectric signal generated by the brachial artery pulsation. In some embodiments of the present application, the piezoelectric device 203 can be implemented based on a piezoelectric sensor device. As mentioned above, in the process of measuring blood pressure, the airbag 201 is first inflated and the cuff 200 is pressurized, thereby blocking the blood flow in the upper arm artery. After that, the gas in the airbag 201 is slowly released. In the process of the airbag 201 releasing gas, the piezoelectric device 203 will detect the pressure of the upper arm on the cuff 200 as the brachial artery pulsates and generates a corresponding piezoelectric signal.

[0099] The signal output device 210 is used to output the pressure signal and the piezoelectric signal.

[0100] In some embodiments of the present application, the signal output device 210 can be implemented as a short-range communication module such as a Bluetooth module to output the pressure signal detected by the pressure detection device 202 and the piezoelectric signal detected by the piezoelectric device 203 to a matching signal processing device through a non-contact communication connection.

[0101] In other embodiments of the present application, the signal output device 210 may also be a signal output interface to output the pressure signal detected by the pressure detection device 202 and the piezoelectric signal detected by the piezoelectric device 203 to a signal processing device connected to the blood pressure measurement device via a contactless communication connection for subsequent signal processing to obtain the user's blood pressure value. The signal processing device may be a smart terminal with signal processing capabilities.

[0102] In other embodiments of the present application, Figure 3 As shown, the blood pressure measurement device includes: a cuff 300 and a signal processing device 310, as well as an air bag 301, a pressure detection device 302, and a piezoelectric device 303 provided in the cuff 300. The pressure detection device 302 and the piezoelectric device 303 are respectively connected to the signal processing device 310 for communication.

[0103] The pressure detection device 302 is configured to collect the pressure signal in the airbag 301;

[0104] The piezoelectric device 303 is provided on the side of the cuff 300 that fits against the user's arm and is configured to collect the piezoelectric signal generated by the brachial artery pulsation;

[0105] The signal processing device 310 is used to obtain the pressure signal and piezoelectric signal collected by the blood pressure measuring device during the blood pressure measurement process of the current user; perform signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation; perform feature extraction on the sequence through a pre-trained feature extraction network to obtain the current blood pressure feature; determine the user category matching the current user based on the current blood pressure feature and the registered blood pressure feature of the registered user of the blood pressure measuring device, wherein the user category includes: specifying the registered user or the guest user.

[0106] The specific implementation method of performing signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation is described below and is not repeated here.

[0107] The sequence is subjected to feature extraction through a pre-trained feature extraction network to obtain current blood pressure features; based on the current blood pressure features and the registered blood pressure features of the registered user of the blood pressure measurement device, the specific implementation method of determining the user category matching the current user is described below and will not be repeated here.

[0108] The structure of the airbag 301 is Figure 2 The structure and working principle of the middle airbag 201 are the same and will not be described again here.

[0109] The setting position, specific implementation method, and principle of collecting pressure signals of the pressure detection device 302 are the same as those of the aforementioned pressure detection device 202 and will not be repeated here.

[0110] The installation position, specific implementation method, and principle of collecting piezoelectric signals of the piezoelectric device 303 are the same as those of the aforementioned piezoelectric device 203 and will not be described in detail here.

[0111] Figure 3 The blood pressure measuring device described in Figure 2 The difference between the blood pressure measuring devices shown in Figure 3 The blood pressure measurement device shown in the figure is provided with a signal processing device 310, which can perform signal processing on the pressure signal detected by the pressure detection device 302 and the piezoelectric signal detected by the piezoelectric device 303 to obtain the user category matched by the current user.

[0112] exist Figure 2 and Figure 3 In the blood pressure measurement device shown, the piezoelectric device 203 includes a piezoelectric sensor integrated with the cuff and disposed on the inside of the cuff. When measuring a user's blood pressure, the piezoelectric sensor is placed in close proximity to the brachial artery in the user's upper arm. The inside of the cuff refers to the side closest to the user's arm skin when measuring blood pressure. The piezoelectric sensor can be a piezoelectric sheet. The piezoelectric sensor can be attached to the cuff by gluing or sewing.

[0113] The piezoelectric device 203 includes at least two piezoelectric sensor devices, and the piezoelectric signal is the superimposed piezoelectric signal of the at least two piezoelectric sensor devices. For example, the piezoelectric signal can be the sum of the amplitudes of the piezoelectric signals collected by the two piezoelectric sensor devices at the same time.

[0114] By adopting a dual piezoelectric sensor device structure and superimposing the piezoelectric signals collected by the two piezoelectric sensor devices as the final piezoelectric signal for identification, not only can the piezoelectric signal collection capability be improved, but also, by setting the two piezoelectric sensor devices close to the brachial artery in the arm of the user being measured, when measuring the user's blood pressure, both piezoelectric sensor devices can detect the signal of the brachial artery, which can improve the tolerance of the position where the user wears the cuff, making the blood pressure measurement equipment easier to use.

[0115] In the process of measuring blood pressure, when the blood in the heart returns, the brachial artery will contract. As the blood in the heart increases, the brachial artery will contract to its narrowest point. When the heart pumps blood out, the brachial artery will expand to its widest point as the blood in the blood vessel increases. The widest and narrowest blood pressure values are the person's high and low blood pressure. In the process of measuring the user's blood pressure using the blood pressure measuring device disclosed in the embodiment of the present application, the cuff is tied to the outside of the brachial artery. When the airbag is inflated, the pressure inside the airbag gradually increases. When the pressure is greater than a certain pressure value, the brachial artery will be closed due to the pressure. Subsequently, the airbag is deflated, and the pressure inside the airbag gradually decreases. At this time, the brachial artery will open, and blood will flow through the brachial artery at the cuff binding position, generating pressure on the airbag and the cuff. The blood pressure measuring device disclosed in the embodiment of the present application detects the change in pressure in the airbag by a pressure detection device arranged in the airbag of the cuff to obtain a pressure signal, and detects the pressure of the brachial artery on the cuff and the airbag by a piezoelectric device to obtain a piezoelectric signal.

[0116] In the process of measuring blood pressure using the blood pressure measuring device disclosed in the embodiment of the present application, the pressure signal detected by the pressure detection device of the blood pressure measuring device is as follows: Figure 4 As shown. Among them, the pressure signal is an oscillation wave, Figure 4 The curve in reflects the changing trend of the shock wave baseline, that is, the changing trend of the pressure value.

[0117] In the process of measuring blood pressure using the blood pressure measuring device disclosed in the embodiment of the present application, the piezoelectric signal detected by the piezoelectric device of the blood pressure measuring device is as follows: Figure 5 As shown. Figure 5 It can be seen that the piezoelectric signal is a pulse signal with a varying peak value. The pulse generation period corresponds to the brachial artery pulsation, and the peak value of the pulse varies with the measurement time.

[0118] The pressure signal and the piezoelectric signal can accurately reflect the changes in the user's blood pressure value. In the embodiment of the present application, the user's blood pressure characteristics are extracted by comprehensively analyzing the pressure signal and the piezoelectric signal.

[0119] Step 120 : Process the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation.

[0120] From the above analysis, it can be seen that the changes in the pressure signal and the piezoelectric signal are signals reflecting the frequency and amplitude of the brachial artery pulsation. Next, the pressure signal and the piezoelectric signal are further processed to obtain a sequence of signal characteristics corresponding to each brachial artery pulsation.

[0121] In some embodiments of the present application, the signal characteristics include: pressure wave difference, first piezoelectric signal characteristics, second piezoelectric signal characteristics, and acquisition time characteristics corresponding to brachial artery pulsation. The signal processing of the pressure signal and the piezoelectric signal to obtain a sequence of signal characteristics corresponding to each brachial artery pulsation includes: sub-steps 1201 to 1207.

[0122] Sub-step 1201: performing band-pass filtering on the pressure signal to obtain an oscillation wave signal within a preset frequency range.

[0123] The preset frequency range is determined based on statistical data of a person's heart rate range. For example, if a person's heart rate range is 40-200 beats per minute, the preset frequency range can be set to [0.67, 3.3].

[0124] The bandpass filter will filter out the frequency signals outside the bandpass frequency in the original signal, thereby changing the original signal. In the embodiment of the present application, a bandpass filter with a frequency range of [0.67, 3.3] can be used to bandpass filter the pressure signal detected by the pressure detection device to filter out the interference signals outside the frequency range. The resulting shock wave signal is as follows: Figure 6 The band-pass filter may be a Butterworth band-pass filter.

[0125] The shock wave signal is a function of the acquisition time, the independent variable is the acquisition time, and the dependent variable is the pressure change value.

[0126] The specific implementation method of band-pass filtering the pressure signal to obtain an oscillation wave signal within a preset frequency range can refer to the existing technology and will not be described in detail in the embodiments of this application.

[0127] Sub-step 1202: determining the acquisition time of the pressure signal corresponding to each peak value of the shock wave signal.

[0128] Sub-step 1203: determining the acquisition time feature corresponding to the corresponding brachial artery pulsation according to the acquisition time.

[0129] Afterwards, the acquisition time of the pressure signal corresponding to each peak of the oscillation wave signal is further obtained. According to the detection principle of the aforementioned pressure signal, each period of the oscillation wave signal corresponds to a brachial artery pulsation, and the time difference between adjacent peaks of the oscillation wave signal is the period of the brachial artery pulsation. Therefore, the acquisition time of the pressure signal corresponding to each peak can be used as a feature to identify a brachial artery pulsation. In an embodiment of the present application, the acquisition time can be used as the acquisition time feature corresponding to the corresponding brachial artery pulsation; the acquisition time sequence number corresponding to each brachial artery pulsation can also be determined based on the acquisition time, and the time sequence number can be used as the acquisition time feature corresponding to the corresponding brachial artery pulsation.

[0130] Next, other features corresponding to each brachial artery pulse are determined separately.

[0131] Sub-step 1204 , determining the pressure wave difference corresponding to the acquisition time according to the change amplitude of the pressure signal corresponding to the acquisition time in the shock wave signal.

[0132] In the embodiment of the present application, the feature extracted from the pressure signal corresponding to each brachial artery pulsation is the pressure wave difference of the oscillatory wave signal. The pressure wave difference can reflect the pressure change amplitude generated by a brachial artery pulsation.

[0133] Furthermore, determining the pressure wave difference corresponding to the acquisition time based on the amplitude of the pressure signal change in the oscillation wave signal corresponding to the acquisition time includes: determining the pressure wave difference corresponding to the acquisition time based on the signal maximum and signal minimum in the oscillation wave signal corresponding to the acquisition time. For example, for each acquisition time determined in the aforementioned step, the signal maximum and signal minimum corresponding to the oscillation wave signal within the brachial artery pulsation cycle corresponding to each acquisition time can be determined based on the oscillation wave function, that is, the peak value and trough value within the oscillation wave signal cycle corresponding to each acquisition time. Thereafter, the difference obtained by subtracting the signal minimum from the signal maximum is used as the pressure wave difference corresponding to the acquisition time.

[0134] Sub-step 1205 , determining a first piezoelectric signal feature corresponding to the acquisition time based on a first frequency band piezoelectric signal obtained by performing bandpass filtering on the piezoelectric signal.

[0135] The piezoelectric signal (e.g., a dual piezoelectric sensor signal) refers to a superimposed signal of the original piezoelectric signal collected from the dual piezoelectric patch. The piezoelectric bandpass can better reflect the characteristics of the high voltage value. Therefore, in some embodiments of the present application, the first piezoelectric signal characteristic corresponding to the brachial artery pulsation is determined based on the first frequency band piezoelectric signal obtained by bandpass filtering the piezoelectric signal.

[0136] In some embodiments of the present application, a first piezoelectric signal characteristic corresponding to the corresponding acquisition time is determined based on a first frequency band piezoelectric signal obtained after band-pass filtering the piezoelectric signal, including: band-pass filtering the piezoelectric signal based on a first preset frequency bandwidth to obtain a first frequency band piezoelectric signal, and then determining the first piezoelectric signal characteristic corresponding to the corresponding acquisition time based on the piezoelectric signal strength corresponding to the first frequency band piezoelectric signal and each acquisition time.

[0137] The first preset frequency bandwidth is selected based on the characteristics of the high and low voltage signals. A piezoelectric signal with a frequency of [20, 40] has a stronger ability to express the characteristics of the piezoelectric signal generated by arterial dilation. Therefore, the bandwidth of the bandpass filter, i.e., the first preset frequency bandwidth, can be set to [20, 40]. The bandpass filter can be a Butterworth bandpass filter.

[0138] After the piezoelectric signal is band-pass filtered by a band-pass filter, the following is obtained: Figure 6 The first frequency band piezoelectric signal is shown in the bottom row of waveforms. Figure 6As shown, the first frequency band piezoelectric signal has a piezoelectric signal intensity distribution corresponding to the brachial artery fluctuation period. Next, the first piezoelectric signal characteristics corresponding to the corresponding acquisition time can be determined based on the piezoelectric signal intensity corresponding to each acquisition time determined in the above step.

[0139] In some embodiments of the present application, the first piezoelectric signal feature corresponding to the corresponding acquisition time is determined based on the piezoelectric signal strength corresponding to the first-frequency-band piezoelectric signal and each of the acquisition times, including: for each of the acquisition times, taking the average value of the piezoelectric signal strength collected at a preset number of acquisition time points within the time period corresponding to the acquisition time and the first-frequency-band piezoelectric signal as the first piezoelectric signal feature corresponding to the acquisition time; or, for each of the acquisition times, taking the root mean square value of the piezoelectric signal strength collected at a preset number of acquisition time points within the time period corresponding to the acquisition time and the first piezoelectric signal as the first piezoelectric signal feature corresponding to the acquisition time.

[0140] The preset number of acquisition time points may be a plurality of time points with preset time intervals within the brachial artery pulsation cycle corresponding to the acquisition time.

[0141] Sub-step 1206, determining a second piezoelectric signal characteristic corresponding to the corresponding acquisition time based on a second frequency band piezoelectric signal obtained after performing segmented bandpass filtering on the piezoelectric signal, wherein the frequency of the first frequency band piezoelectric signal is lower than that of the second frequency band piezoelectric signal.

[0142] Piezoelectric segmented bandpass can better reflect the characteristics of low voltage values. Therefore, in some embodiments of the present application, the second piezoelectric signal characteristics corresponding to the brachial artery pulsation are determined based on the second frequency band piezoelectric signal obtained after segmented bandpass filtering of the piezoelectric signal.

[0143] In some embodiments of the present application, a second piezoelectric signal characteristic corresponding to the corresponding acquisition time is determined based on a second frequency band piezoelectric signal obtained after performing segmented bandpass filtering on the piezoelectric signal, including: performing segmented bandpass filtering on the piezoelectric signal based on a second preset frequency bandwidth to obtain a second frequency band piezoelectric signal, and then determining the second piezoelectric signal characteristic corresponding to the corresponding acquisition time based on the piezoelectric signal strength corresponding to the second frequency band piezoelectric signal and each acquisition time.

[0144] The second preset frequency bandwidth may be, for example, a segmented bandpass bandwidth of [40, 45], [55, 95], [105, 145], or [155, 195] to filter power frequency interference.

[0145] By observing the piezoelectric signals, the inventors found that the piezoelectric signal with a frequency of [20, 40] has a stronger ability to express the characteristics of the piezoelectric signal generated when the arterial blood vessels expand, and the piezoelectric signal with a frequency of [40, 200] has a stronger ability to express the characteristics of the piezoelectric signal generated when the arterial blood vessels contract. Therefore, by performing the above-mentioned band-pass filtering and segmented band-pass filtering on the piezoelectric signal, the first frequency band piezoelectric signal and the second frequency band piezoelectric signal obtained respectively can effectively filter out interference and retain the piezoelectric signal used to express the brachial artery pulsation characteristics.

[0146] Sub-step 1207 , generating a sequence of signal features by combining the acquisition time features, the pressure wave difference, the first piezoelectric signal features, and the second piezoelectric signal features corresponding to each acquisition time in the order of the acquisition times.

[0147] After obtaining the acquisition time feature, the pressure wave difference, the first piezoelectric signal feature, and the second piezoelectric signal feature corresponding to each acquisition time, the aforementioned features are arranged from front to back in the order of the acquisition times to generate the sequence of signal features. For example, for a certain acquisition time, the acquisition time feature, the pressure wave difference, the first piezoelectric signal feature, and the second piezoelectric signal feature corresponding to the acquisition time can be collectively used as a four-dimensional signal feature corresponding to the acquisition time; thereafter, the signal features corresponding to N acquisition times are arranged in order in the order of the acquisition times to obtain a sequence of signal features of length N×4, where N is an integer greater than 1.

[0148] Step 130 : extracting features from the sequence using a pre-trained feature extraction network to obtain current blood pressure features.

[0149] The sequence of signal features obtained in the above steps is the time series feature corresponding to the brachial artery pulsation signal. Next, the sequence is input into a feature extraction network pre-trained based on a time series model to extract features from the sequence.

[0150] In some embodiments of the present application, the feature extraction network includes: a bidirectional long short-term memory network and a feature conversion layer. The feature extraction of the sequence is performed through the pre-trained feature extraction network to obtain the current blood pressure feature, including: inputting the sequence into the bidirectional long short-term memory network, and performing feature encoding on each signal feature in the sequence through the bidirectional long short-term memory network to obtain a hidden layer vector corresponding to each signal feature; calling the feature conversion layer to perform an arithmetic mean operation on the hidden layer vector of the signal feature to obtain the current blood pressure feature.

[0151] In some embodiments of the present application, the feature extraction network is part of a pre-trained user recognition model. The user recognition model includes: the feature extraction network and a fully connected layer. In the training stage of the user recognition model, after the sequence is input into the user recognition model, the sequence of input signal features is first feature-encoded by the feature extraction network to obtain a hidden vector corresponding to each signal feature; then, the feature conversion layer is called to perform an arithmetic mean operation on the hidden vector of the signal feature to obtain the blood pressure feature corresponding to the sequence. Afterwards, the blood pressure feature is input into the fully connected layer, and each blood pressure feature is classified and mapped by the fully connected layer to obtain the user classification result corresponding to the sequence.

[0152] The feature extraction process of the feature extraction network is the same in the training phase and the application phase.

[0153] like Figure 7 As shown, the sequence of signal features is X0, X1, X2,…, X n-1 ,X n , where X i It consists of the four-dimensional data of the acquisition time feature, pressure wave difference, the first piezoelectric signal feature, and the second piezoelectric signal. Each signal feature in the sequence is input into the feature extraction network in turn, and each X is extracted by the bidirectional long short-term memory network. i Feature encoding is performed to obtain a hidden layer vector corresponding to each of the signal features. Afterwards, each hidden layer vector is sent to the feature conversion layer for feature conversion. The feature conversion layer performs an arithmetic mean operation on the hidden layer vectors of the signal features to obtain the blood pressure feature corresponding to the sequence.

[0154] By adopting a bidirectional long short-term memory network to construct a feature extraction network, the extracted signal features can integrate the information of the pressure signal and piezoelectric signal collected at various time points during the blood pressure measurement process, further improving the feature expression ability of the blood pressure features extracted by the feature extraction network.

[0155] In some embodiments of the present application, the feature extraction network can also be constructed based on neural networks of other structures, which are not listed one by one in this article.

[0156] Step 140 : Determine the user category that matches the current user based on the current blood pressure characteristics and the registered blood pressure characteristics of the registered user of the blood pressure measurement device, wherein the user category includes: specifying the registered user or the guest user.

[0157] After obtaining the current blood pressure characteristics, by matching the current blood pressure characteristics with the registered blood pressure characteristics of each registered user in the blood pressure measurement device, it can be determined whether the current user is a guest user or a registered user.

[0158] In some embodiments of the present application, determining the user category that matches the current user based on the current blood pressure characteristics and the registered blood pressure characteristics of the registered user of the blood pressure measuring device includes: obtaining the similarity distance between the registered blood pressure characteristics of the registered user of the blood pressure measuring device and the current blood pressure characteristics; determining the registered user to whom the registered blood pressure characteristics corresponding to the minimum value of the similarity distance belong as a candidate registered user; and determining the user category that matches the current user based on the relationship between the similarity distance threshold corresponding to the candidate registered user and the minimum value of the similarity distance.

[0159] For example, a blood pressure measurement device includes registered users A, B, and C. The blood pressure measurement device stores the registered blood pressure features of registered users A, B, and C. After obtaining the current blood pressure feature of the current user Y of the blood pressure measurement device, the similarity distance D between the blood pressure feature of the current user Y (i.e., the current blood pressure feature) and the registered blood pressure features of registered users A, B, and C is calculated. A 、D B and D C , and obtain three similarity distances. Then, the registered user corresponding to the smallest similarity distance among the three similarity distances is selected as the candidate registered user. For example, the similarity distance D between the blood pressure feature of the current user Y and the registered blood pressure feature of the registered user A is A Less than the similarity distance D between the registered blood pressure features of registered users B and C B and D C , then the registered user A is determined as the candidate registered user. That is, the current user is most likely to be the registered user A.

[0160] In some embodiments of the present application, after a user completes registration on a blood pressure measurement device, the blood pressure measurement device also stores a similarity distance threshold corresponding to the registered user. The similarity distance threshold is used to determine whether a user is the same user as the registered user. For example, for registered users A, B, and C, the blood pressure measurement device also stores the corresponding similarity distance threshold L A , L B and L C .

[0161] In some embodiments of the present application, the user category matched by the current user is determined based on the size relationship between the similarity distance threshold corresponding to the candidate registered user and the minimum similarity distance value, including: adjusting the similarity distance threshold according to a preset ratio value to obtain a reference threshold, wherein the preset ratio value is greater than or equal to 1; in response to the minimum similarity distance value being less than or equal to the reference threshold, determining that the user category matched by the current user is the candidate registered user; in response to the minimum similarity distance value being greater than the reference threshold, determining that the current user matches a visitor user.

[0162] Optionally, the preset ratio value is a value greater than 1. The similarity distance threshold L corresponding to the candidate registered user A is used. A For example, if the preset ratio value is 1.3, you can set 1.3×L A As a reference threshold. Further, if the similarity distance D between the blood pressure feature of the current user Y and the registered blood pressure feature of the registered user A is A ≤1.3×L A , it can be considered that the user category matched by the current user Y is the candidate registered user, that is, the current user Y is the registered user A; if the similarity distance D between the blood pressure feature of the current user Y and the registered blood pressure feature of the registered user A A >1.3×L A , it can be considered that the current user Y and the candidate registered user are not the same user, that is, the current user Y is a guest user.

[0163] In some embodiments of the present application, the registered blood pressure characteristics of the registered user are determined in advance by the following method: in the user registration mode, the pressure signal and piezoelectric signal collected by the blood pressure measuring device for N blood pressure measurement processes of the registered user are obtained, and the blood pressure characteristics corresponding to each blood pressure measurement process are obtained based on the pressure signal and the piezoelectric signal; the registered blood pressure characteristics of the registered user are obtained based on the average characteristics of the blood pressure characteristics corresponding to the N blood pressure measurement processes, where N is an integer greater than 1.

[0164] Optionally, the user registration mode can be started by pressing a button or in other ways.

[0165] Optionally, the blood pressure characteristics corresponding to each blood pressure measurement process are obtained based on the pressure signal and the piezoelectric signal, including: performing signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal characteristics corresponding to each brachial artery pulsation; performing feature extraction on the sequence through a pre-trained feature extraction network to obtain the blood pressure characteristics corresponding to the current blood pressure measurement process.

[0166] Taking the registration process of registered user A as an example, registered user A first measures blood pressure N times (e.g., 5 times). During each blood pressure measurement, the blood pressure measurement device collects pressure signals and piezoelectric signals, and processes the pressure signals and piezoelectric signals collected for each blood pressure measurement according to the aforementioned method to obtain a sequence of signal features corresponding to each brachial artery pulsation. Afterwards, the sequence of signal features corresponding to each blood pressure measurement is subjected to feature extraction by a pre-trained feature extraction network, and the blood pressure features corresponding to the corresponding blood pressure measurement processes are obtained. In this way, N groups of blood pressure features can be obtained. Then, the average feature Fa of these N groups of blood pressure features is calculated as the registered blood pressure feature of the registered user A.

[0167] In some embodiments of the present application, the similarity distance threshold corresponding to the candidate registered user is determined in advance by the following method: the similarity distance threshold corresponding to the candidate registered user is obtained based on the maximum similarity distance between the blood pressure feature and the average feature corresponding to the N blood pressure measurement processes of the candidate registered user.

[0168] Still taking the registration process of registered user A as an example, after obtaining the average feature Fa of registered user A, the Euclidean distance between the average feature Fa and N groups of blood pressure features is calculated to obtain N Euclidean distances, and the largest Euclidean distance is used as the similarity distance threshold corresponding to registered user A.

[0169] In order to make this solution clearer, the training process of the feature extraction network is further explained below.

[0170] like Figure 8 As shown, before obtaining the pressure signal and piezoelectric signal collected by the blood pressure measurement device during the blood pressure measurement process of the current user, the method further includes: step 100 and step 102.

[0171] Step 100: Obtain several training samples.

[0172] Step 102: training a feature extraction network based on the plurality of training samples.

[0173] As mentioned above, the feature extraction network is a part of the user identification model. Therefore, the training process of the user identification model is the training process of the feature extraction network.

[0174] In some embodiments of the present application, the feature extraction network is obtained by training a user recognition model based on a blood pressure signal, wherein the user recognition model based on a blood pressure signal includes: the feature extraction network and a fully connected layer, and the user recognition model based on a blood pressure signal is trained by the following method: obtaining a number of training samples, wherein the sample data of each training sample is: a sequence of signal features obtained after processing the pressure signal and the piezoelectric signal collected based on a single blood pressure measurement, and the sample label is: the true value of the user category; for each of the training samples, the sample data is feature extracted by the feature extraction network, and the hidden layer vector obtained by the feature extraction is used as the blood pressure feature corresponding to the training sample; the blood pressure feature is classified and mapped by the fully connected layer to obtain the user category prediction value corresponding to the training sample; the model loss of the user recognition model based on the blood pressure signal is calculated according to the user category prediction value and the sample label; and the user recognition model based on the blood pressure signal is iteratively trained by optimizing the model loss to obtain the feature extraction network.

[0175] Among them, each training sample used to train the user recognition model can be a sequence of signal features obtained after processing the pressure signal and piezoelectric signal collected by the blood pressure measurement device disclosed in the embodiment of the present application during a blood pressure measurement of the user, and the true value of the user category corresponding to each sequence is set as the sample label corresponding to the sequence.

[0176] During the process of training the user identification model, after the sequence of signal features in each training sample is input into the user identification model, first, the sequence is subjected to feature extraction through the feature extraction network, and the hidden layer vector obtained by the feature extraction is used as the blood pressure feature corresponding to the training sample; then, the blood pressure feature is classified and mapped through the fully connected layer to obtain the user category prediction value corresponding to the training sample.

[0177] Next, the model loss of the user identification model is calculated based on the user category prediction value and the sample label of the training sample. In some embodiments of the present application, the loss function of the user identification model selects Softmaxloss and Centerloss. Softmaxloss is used to separate different categories, and centerloss is used to compress the same category. Adding Center loss on the basis of Softmaxloss can increase the inter-class distance and reduce the intra-class distance of the classified samples in the feature space, making the sample features separable in the Euclidean space and improving the expressive power of the features extracted by the feature extraction network.

[0178] Next, the user identification model is iteratively trained with the goal of optimizing the model loss until the model loss converges, completing the training. The network parameters of the user identification model obtained at this point are optimal, meaning that the network parameters of the feature extraction network are also optimal. This completes the training of the feature extraction network.

[0179] The user classification method for a blood pressure measuring device disclosed in an embodiment of the present application obtains a pressure signal and a piezoelectric signal collected during the blood pressure measurement process of the current user by the blood pressure measuring device; performs signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation; extracts features from the sequence through a pre-trained feature extraction network to obtain current blood pressure features; determines a user category matching the current user based on the current blood pressure features and the registered blood pressure features of the registered user of the blood pressure measuring device, wherein the user category includes: specifying the registered user or the guest user, thereby realizing the distinction between different measured persons based on blood pressure features, solving the problems that the number of user categories of the blood pressure measuring device is limited, the measurement data is easily interfered with by the measurement data of other users, and the use is cumbersome.

[0180] The user classification method for the blood pressure measuring device disclosed in the embodiment of the present application extracts the blood pressure characteristics of the person being measured, and compares the extracted blood pressure characteristics with the blood pressure characteristics extracted when the user registers to identify the user's identity. The above-mentioned blood pressure characteristics are deep-level stable characteristics extracted by a neural network during the user's blood pressure measurement process. Regardless of whether the user's blood pressure measurement is normal or abnormal, the user's identity can be accurately identified through the deep-level stable characteristics, that is, the blood pressure characteristics obtained after signal processing of the pressure signal and piezoelectric signal collected during the user's blood pressure measurement process. Thereby, the blood pressure data of other users does not interfere with the blood pressure data of the registered user, and the number of registered users is not limited. In addition, the blood pressure measuring device that adopts the user classification method for the blood pressure measuring device disclosed in the embodiment of the present application does not need to manually switch users when performing blood pressure measurement, and the operation is more convenient.

[0181] The present application also discloses a user classification device for a blood pressure measurement device, such as Figure 9 As shown, the device includes:

[0182] A blood pressure measurement signal acquisition module 910 is configured to acquire a pressure signal and a piezoelectric signal collected by a blood pressure measurement device during a blood pressure measurement process for a current user;

[0183] a signal feature sequence acquisition module 920 for performing signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation;

[0184] The current blood pressure feature acquisition module 930 is used to extract features from the sequence using a pre-trained feature extraction network to obtain current blood pressure features;

[0185] The user classification module 940 is configured to determine a user category that matches the current user based on the current blood pressure characteristics and the registered blood pressure characteristics of the registered user of the blood pressure measurement device, wherein the user category includes specifying the registered user or the guest user.

[0186] In some embodiments of the present application, the user separation module 940 is further configured to:

[0187] Obtaining a similarity distance between a registered blood pressure feature of a registered user of the blood pressure measurement device and the current blood pressure feature;

[0188] Determine the registered user to whom the registered blood pressure feature corresponding to the minimum similarity distance belongs as a candidate registered user;

[0189] The user category that the current user matches is determined based on the magnitude relationship between the similarity distance threshold corresponding to the candidate registered user and the similarity distance minimum value.

[0190] In some embodiments of the present application, determining the user category that matches the current user based on the relationship between the similarity distance threshold corresponding to the candidate registered user and the minimum similarity distance value includes:

[0191] Adjusting the similarity distance threshold according to a preset ratio value to obtain a reference threshold, wherein the preset ratio value is greater than or equal to 1;

[0192] In response to the minimum similarity distance being less than or equal to the reference threshold, determining that the user category matched by the current user is the candidate registered user;

[0193] In response to the minimum similarity distance being greater than the reference threshold, it is determined that the current user matches the guest user.

[0194] In some embodiments of the present application, the registered blood pressure characteristics of the registered user are pre-determined by the following method:

[0195] In the user registration mode, obtaining pressure signals and piezoelectric signals collected by the blood pressure measurement device during N blood pressure measurement processes of the registered user, and obtaining blood pressure characteristics corresponding to each blood pressure measurement process based on the pressure signals and the piezoelectric signals;

[0196] The registered blood pressure feature of the registered user is obtained according to the average feature of the blood pressure features corresponding to N blood pressure measurement processes, where N is an integer greater than 1.

[0197] In some embodiments of the present application, the similarity distance threshold corresponding to the candidate registered user is pre-determined by the following method:

[0198] A similarity distance threshold corresponding to the candidate registered user is obtained according to the maximum similarity distance between the blood pressure feature corresponding to the N blood pressure measurement processes of the candidate registered user and the average feature.

[0199] In some embodiments of the present application, the pressure signal is a pressure signal within the airbag of the cuff of the blood pressure measurement device, and the piezoelectric signal is collected by a piezoelectric device provided on the side of the cuff that contacts the brachial artery of the current user. The signal characteristics include: a pressure wave difference, a first piezoelectric signal characteristic, a second piezoelectric signal characteristic, and an acquisition time characteristic corresponding to the brachial artery pulsation. The signal characteristic sequence acquisition module 920 is further used to:

[0200] Band-pass filtering is performed on the pressure signal to obtain an oscillation wave signal within a preset frequency range;

[0201] Determining the acquisition time of the pressure signal corresponding to each peak value of the shock wave signal;

[0202] Determining, based on the acquisition time, the acquisition time feature corresponding to the corresponding brachial artery pulsation;

[0203] determining a pressure wave difference corresponding to the acquisition time according to a change amplitude of a pressure signal in the shock wave signal corresponding to the acquisition time;

[0204] determining a first piezoelectric signal feature corresponding to the acquisition time according to a first frequency band piezoelectric signal obtained by performing bandpass filtering on the piezoelectric signal;

[0205] determining a second piezoelectric signal feature corresponding to the acquisition time based on a second frequency band piezoelectric signal obtained by performing segmented bandpass filtering on the piezoelectric signal, wherein the frequency of the first frequency band piezoelectric signal is lower than that of the second frequency band piezoelectric signal;

[0206] The acquisition time feature, the pressure wave difference, the first piezoelectric signal feature, and the second piezoelectric signal feature corresponding to each acquisition time are used in the order of the acquisition time to generate a sequence of the signal features.

[0207] In some embodiments of the present application, the feature extraction network includes: a bidirectional long short-term memory network, a feature conversion layer, and the current blood pressure feature acquisition module 930 is further used to:

[0208] Inputting the sequence into the bidirectional long short-term memory network, performing feature encoding on each signal feature in the sequence through the bidirectional long short-term memory network, and obtaining a hidden layer vector corresponding to each signal feature;

[0209] The feature conversion layer is called to perform an arithmetic mean operation on the hidden layer vector of the signal feature to obtain the current blood pressure feature.

[0210] Optionally, the feature extraction network is obtained by training a user identification model based on a blood pressure signal, wherein the user identification model based on a blood pressure signal comprises: the feature extraction network and a fully connected layer, and the user identification model based on a blood pressure signal is trained by the following method:

[0211] Acquire a number of training samples, wherein the sample data of each training sample is: a sequence of signal features obtained by processing the pressure signal and piezoelectric signal collected based on a single blood pressure measurement, and the sample label is: the true value of the user category;

[0212] For each of the training samples, extracting features from the sample data using the feature extraction network, and using the hidden layer vector obtained by the feature extraction as the blood pressure feature corresponding to the training sample;

[0213] Classify and map the blood pressure features through the fully connected layer to obtain a user category prediction value corresponding to the training sample;

[0214] Calculating a model loss of the user recognition model based on the blood pressure signal according to the user category prediction value and the sample label;

[0215] The user recognition model based on the blood pressure signal is iteratively trained by optimizing the model loss to obtain the feature extraction network.

[0216] The user classification device of the blood pressure measuring device disclosed in the embodiment of the present application is used to implement the user classification method of the blood pressure measuring device described in the embodiment of the present application. The specific implementation methods of each module of the device will not be repeated here. Please refer to the specific implementation methods of the corresponding steps in the method embodiment.

[0217] The user classification device of the blood pressure measuring device disclosed in the embodiment of the present application obtains the pressure signal and the piezoelectric signal collected during the blood pressure measurement process of the current user by the blood pressure measuring device; performs signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation; extracts features from the sequence through a pre-trained feature extraction network to obtain the current blood pressure feature; determines the user category matching the current user based on the current blood pressure feature and the registered blood pressure feature of the registered user of the blood pressure measuring device, wherein the user category includes: specifying the registered user or the guest user, thereby realizing the distinction between different measured persons based on blood pressure features, solving the problems that the number of user categories of the blood pressure measuring device is limited, the measurement data is easily interfered with by the measurement data of other users, and the use is cumbersome.

[0218] The user classification device of the blood pressure measuring device disclosed in the embodiment of the present application extracts the blood pressure characteristics of the person being measured, and compares the extracted blood pressure characteristics with the blood pressure characteristics extracted when the user is registered to identify the user's identity. The above-mentioned blood pressure characteristics are deep-level stable characteristics extracted by a neural network during the user's blood pressure measurement process. Regardless of whether the user's blood pressure measurement is normal or abnormal, the user's identity can be accurately identified through the deep-level stable characteristics, that is, the blood pressure characteristics obtained after signal processing of the pressure signal and piezoelectric signal collected during the user's blood pressure measurement process. Thereby, the blood pressure data of other users does not interfere with the blood pressure data of the registered user, and the number of registered users is not limited. In addition, the blood pressure measuring device that adopts the user classification method of the blood pressure measuring device disclosed in the embodiment of the present application does not need to manually switch users when performing blood pressure measurement, and the operation is more convenient.

[0219] Each embodiment in this specification is described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between the various embodiments can be referred to in conjunction with each other. For the device embodiments, since they are generally similar to the method embodiments, their description is relatively simple, and for relevant parts, reference can be made to the description of the method embodiments.

[0220] The above is a detailed introduction to the user classification method and device for a blood pressure measurement device provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, based on the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0221] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0222] The various component embodiments of the present application can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. It will be appreciated by those skilled in the art that a microprocessor or digital signal processor (DSP) can be used in practice to implement some or all of the functions of some or all of the components in the electronic device according to the embodiment of the present application. The application can also be implemented as a device or apparatus program (for example, a computer program and a computer program product) for performing a part or all of the methods described herein. Such a program implementing the present application can be stored on a computer-readable medium, or can have the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0223] For example, Figure 10An electronic device that can implement the method according to the present application is shown. The electronic device can be a PC, a mobile terminal, a personal digital assistant, a tablet computer, etc. The electronic device conventionally includes a processor 1010 and a memory 1020, and program code 1030 stored on the memory 1020 and executable on the processor 1010. When the processor 1010 executes the program code 1030, the method described in the above embodiments is implemented. The memory 1020 can be a computer program product or a computer-readable medium. The memory 1020 can be an electronic memory such as a flash memory, an EEPROM (Electrically Erasable Programmable Read-Only Memory), an EPROM, a hard disk, or a ROM. The memory 1020 has a storage space 10201 for program code 1030 of a computer program for executing any of the method steps described above. For example, the storage space 10201 for program code 1030 can include individual computer programs for implementing various steps in the above method. The program code 1030 is computer-readable code. These computer programs can be read from or written into one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks. The computer program includes computer readable code that, when run on an electronic device, causes the electronic device to perform the method according to the above embodiments.

[0224] The embodiment of the present application further discloses a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the steps of the user classification method for a blood pressure measurement device as described in the embodiment of the present application are implemented.

[0225] Such a computer program product may be a computer-readable storage medium having a computer program product. Figure 10 The memory 1020 in the electronic device shown is similarly arranged as a storage segment, storage space, etc. The program code can be compressed and stored in the computer readable storage medium in an appropriate form. The computer readable storage medium is generally as shown in FIG. Figure 11 The portable or fixed storage unit generally includes computer-readable code 1030', which is a code read by a processor and implements the steps of the above-described method when executed by the processor.

[0226] References herein to "one embodiment," "an embodiment," or "one or more embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present application. Furthermore, please note that instances of the phrase "in one embodiment" do not necessarily all refer to the same embodiment.

[0227] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of this description.

[0228] In the claims, any reference signs placed between brackets shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in the claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present application may be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names.

[0229] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for classifying users of a blood pressure measuring device, characterized in that: include: Obtaining a pressure signal and a piezoelectric signal collected by a blood pressure measuring device during a blood pressure measurement process for a current user; performing signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation; Extract features from the sequence using a pre-trained feature extraction network to obtain current blood pressure features; Determining a user category matching the current user based on the current blood pressure characteristics and the registered blood pressure characteristics of the registered user of the blood pressure measuring device, wherein the user category includes: specifying the registered user or the guest user; The pressure signal is a pressure signal within the airbag of the cuff of the blood pressure measurement device, and the piezoelectric signal is collected by a piezoelectric device provided on the side of the cuff that contacts the brachial artery of the current user. The signal characteristics include: pressure wave difference, first piezoelectric signal characteristics, second piezoelectric signal characteristics, and collection time characteristics corresponding to the brachial artery pulsation; The signal processing of the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation includes: Band-pass filtering is performed on the pressure signal to obtain an oscillation wave signal within a preset frequency range; Determining the acquisition time of the pressure signal corresponding to each peak value of the shock wave signal; Determining, based on the acquisition time, the acquisition time feature corresponding to the corresponding brachial artery pulsation; determining a pressure wave difference corresponding to the acquisition time according to a change amplitude of a pressure signal in the shock wave signal corresponding to the acquisition time; determining a first piezoelectric signal feature corresponding to the acquisition time according to a first frequency band piezoelectric signal obtained by performing bandpass filtering on the piezoelectric signal; determining a second piezoelectric signal feature corresponding to the acquisition time based on a second frequency band piezoelectric signal obtained by performing segmented bandpass filtering on the piezoelectric signal, wherein the frequency of the first frequency band piezoelectric signal is lower than that of the second frequency band piezoelectric signal; The acquisition time feature, the pressure wave difference, the first piezoelectric signal feature, and the second piezoelectric signal feature corresponding to each acquisition time are used in the order of the acquisition time to generate a sequence of the signal features.

2. The method according to claim 1, characterized in that The determining, based on the current blood pressure characteristics and the registered blood pressure characteristics of the registered user of the blood pressure measurement device, a user category matching the current user includes: Obtaining a similarity distance between a registered blood pressure feature of a registered user of the blood pressure measurement device and the current blood pressure feature; Determine the registered user to whom the registered blood pressure feature corresponding to the minimum similarity distance belongs as a candidate registered user; The user category that the current user matches is determined based on the magnitude relationship between the similarity distance threshold corresponding to the candidate registered user and the similarity distance minimum value.

3. The method according to claim 2, characterized in that The determining the user category that the current user matches based on the relationship between the similarity distance threshold corresponding to the candidate registered user and the minimum similarity distance value includes: Adjusting the similarity distance threshold according to a preset ratio value to obtain a reference threshold, wherein the preset ratio value is greater than or equal to 1; In response to the minimum similarity distance being less than or equal to the reference threshold, determining that the user category matched by the current user is the candidate registered user; In response to the minimum similarity distance being greater than the reference threshold, it is determined that the current user matches the guest user.

4. The method according to claim 2, characterized in that The registered blood pressure characteristics of the registered user are determined in advance by the following method: In the user registration mode, obtaining pressure signals and piezoelectric signals collected by the blood pressure measurement device during N blood pressure measurement processes of the registered user, and obtaining blood pressure characteristics corresponding to each blood pressure measurement process based on the pressure signals and the piezoelectric signals; The registered blood pressure feature of the registered user is obtained according to the average feature of the blood pressure features corresponding to N blood pressure measurement processes, where N is an integer greater than 1.

5. The method according to claim 4, characterized in that The similarity distance threshold corresponding to the candidate registered user is determined in advance by the following method: A similarity distance threshold corresponding to the candidate registered user is obtained according to the maximum similarity distance between the blood pressure feature corresponding to the N blood pressure measurement processes of the candidate registered user and the average feature.

6. The method according to claim 1, characterized in that The feature extraction network includes: a bidirectional long short-term memory network and a feature conversion layer. The feature extraction network is pre-trained to extract features from the sequence to obtain the current blood pressure features, including: Inputting the sequence into the bidirectional long short-term memory network, performing feature encoding on each signal feature in the sequence through the bidirectional long short-term memory network, and obtaining a hidden layer vector corresponding to each signal feature; The feature conversion layer is called to perform an arithmetic mean operation on the hidden layer vector of the signal feature to obtain the current blood pressure feature.

7. The method according to claim 1, characterized in that The feature extraction network is obtained by training a user identification model based on a blood pressure signal, wherein the user identification model based on a blood pressure signal includes: the feature extraction network and a fully connected layer, and the user identification model based on a blood pressure signal is trained by the following method: Acquire a number of training samples, wherein the sample data of each training sample is: a sequence of signal features obtained by processing the pressure signal and piezoelectric signal collected based on a single blood pressure measurement, and the sample label is: the true value of the user category; For each of the training samples, extracting features from the sample data using the feature extraction network, and using the hidden layer vector obtained by the feature extraction as the blood pressure feature corresponding to the training sample; Classify and map the blood pressure features through the fully connected layer to obtain a user category prediction value corresponding to the training sample; Calculating a model loss of the user recognition model based on the blood pressure signal according to the user category prediction value and the sample label; The user recognition model based on the blood pressure signal is iteratively trained by optimizing the model loss to obtain the feature extraction network.

8. A user classification device for a blood pressure measuring device, which implements the user classification method for a blood pressure measuring device according to any one of claims 1 to 7, characterized in that: include: A blood pressure measurement signal acquisition module is used to obtain a pressure signal and a piezoelectric signal collected by the blood pressure measurement device during the blood pressure measurement process of the current user; a signal feature sequence acquisition module, configured to perform signal processing on the pressure signal and the piezoelectric signal to obtain a sequence of signal features corresponding to each brachial artery pulsation; A current blood pressure feature acquisition module is used to extract features from the sequence using a pre-trained feature extraction network to obtain current blood pressure features; The user classification module is configured to determine a user category that matches the current user based on the current blood pressure characteristics and the registered blood pressure characteristics of the registered user of the blood pressure measuring device, wherein the user category includes: specifying the registered user or the guest user.

9. An electronic device comprising a memory, a processor, and a program code stored in the memory and executable on the processor, wherein: When the processor executes the program code, the user classification method for the blood pressure measurement device according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having program code stored thereon, characterized in that: When the program code is executed by a processor, the steps of the method for classifying users of a blood pressure measurement device according to any one of claims 1 to 7 are implemented.

Images