TW201705903A - Method for identifying user identity by using blood pressure signal - Google Patents

Abstract

A method for identifying user identity by using blood pressure signal, comprises steps of: detecting a blood pressure signal in process of a blood pressure cuff deflating to form a curve with direct-current level and a plurality of waveforms with alternating current level, each of the waveforms corresponding to a pressure value in the curve; when the blood pressure cuff deflating to reach a specific pressure in the curve at intervals of a specific pressure drop, selecting one waveform corresponding to the specific pressure from said waveforms; calculating a first area of a rising band of the selected waveform and a second area of a drop band of the selected waveform to defining a plurality of blood pressure characteristic values; inputting the blood pressure characteristic values into an identification algorithm of user identity, and outputting a recognition result. Using the aforementioned method makes the use of a blood pressure measurement device more convenient, without doing login before measuring blood pressure every time, so as to store the user data automatically.

Description

Method of identifying identity using blood pressure signals

The present invention relates to a method of automatically identifying an identity, and more particularly to a method of identifying an identity using a blood pressure signal.

Due to the aging of society and changes in eating habits, the problems of high blood pressure and high blood fat have plagued many people's lives, so the health management mechanism has gradually received attention. In the health management mechanism, it is an important part to regularly monitor whether the blood pressure signal is abnormal. However, the general sphygmomanometer is still time consuming and inconvenient in operation.

Generally, commercially available sphygmomanometers, after the measurement is completed, directly display the blood pressure value on the machine. When the power of the machine is turned off, no personal measurement record is left. Some sphygmomanometers have a flash memory built in, which allows the user to store the data in the memory after the measurement is completed. After the measurement is completed, the previously measured blood pressure value can be seen. The various recording methods for existing sphygmomanometer products are described as follows:

1. There is no recording function, so it is necessary to manually record the data after the measurement is completed, which is quite laborious and easy to remember.

2. There are multiple sets of memory functions, but the user must coordinate where to remember.

3. Let the user record the measurement results by logging in and measuring. However, logging in every measurement is quite inconvenient, and once the user information is logged in, the blood pressure data will be mistakenly recorded in the records of others. In addition, the elderly have some difficulty in accessing the operation of the sphygmomanometer.

4. User identification is performed by means of the coding module of the user and the detection module of the measuring end, but this method requires only one of the modules to have problems, such as: damage, loss, etc., user and coding module It is impossible to communicate effectively and identify the user between the detection modules.

5. Record blood pressure measurements using RFID or barcodes, but increase costs.

It can be seen from the above description that the inconvenience of the operation of the general sphygmomanometer mainly comes from the recording process of the blood pressure data and the establishment, registration and identification process of the user data. However, since the sphygmomanometer is generally shared by many people, it is necessary to separate each person's record, so the process of establishing, logging in, and identifying the user data is difficult to omit.

In view of this, solving the recording process of measurement data and the convenience of the establishment, registration and identification process of user data can save the operation time of the sphygmomanometer and enhance the public's willingness to use the sphygmomanometer to make the sphygmomanometer product more Competitive and help to establish a sound health management mechanism.

An object of the present invention is to provide a method for identifying an identity by using a blood pressure signal, which automatically recognizes the user's identity according to the characteristic value in the blood pressure signal, and does not need to perform login, typing, and the like before each blood pressure measurement, thereby reducing the automatic The time when the user data is stored improves the convenience of the blood pressure measuring device.

In order to achieve the above object, the present invention provides a method for identifying an identity using a blood pressure signal, which is applied to a blood pressure measuring device. The blood pressure measuring device has a cuff, a pressure sensor and a microcontroller. The method includes the following steps: the pressure sensor detects a change in the internal pressure of the cuff during the deflation process, and forms a blood pressure signal, which includes a constant current signal and an alternating current signal, wherein the direct current signal is formed. A DC curve whose pressure value decreases with time, the AC signal forms a complex AC waveform, wherein each AC waveform has a rising band connected to a falling band and corresponds to a pressure value in the DC curve; The specific pressure difference causes the DC curve to reach a specific pressure value, and the microcontroller selects an AC waveform corresponding to the specific pressure value among the AC waveforms; for the selected AC waveform, the microcontroller calculates the rising wavelength separately The area and the area of the falling band are used to define the complex blood pressure eigenvalues; an identity recognition algorithm is established by a back-transfer neural network, and these blood pressure eigenvalues are input into the identity recognition algorithm, and a recognition result is output.

In one embodiment, the step of selecting the AC waveform by the microcontroller is performed during a period in which the DC curve drops from a high level pressure to a low level pressure, for example, from 150 mmHg to 40 mmHg.

In an embodiment, the method further includes: providing a software program, including an identity recognition algorithm and a mechanism for creating a new member, wherein the mechanism for creating a new member includes: providing an account creation screen for establishing a plurality of accounts, each An account corresponds to a plurality of members in the same family; a screen for creating a new member for creating a member of each member of the family is stored and stored in the account; and after the identity recognition algorithm is executed, the member is generated A recognized member. Next, a selection screen is provided, the selection screen includes the recognized member and other members in the family, wherein the recognized member is represented by a pattern having different transparency; and when the user judges that the recognized member is incorrect, The selection screen provides the user with a correct member to choose from.

In an embodiment, the method further includes: sampling at a sampling frequency in the rising band and the falling band of the selected AC waveform. The area of the rising band and the area of the falling band are calculated as individual squares, and a square area of two areas is obtained, and the square of each area is one of these blood pressure characteristic values.

In an embodiment, the step of establishing an identity recognition algorithm by using the inverse neural network includes: adding a recognition result every time a user measures the blood pressure by the blood pressure measurement device; The result and a known correct data are uploaded to a cloud system; and in the cloud system, the learning process of the reverse neural network is performed by using the newly added identification result and the known correct data to correct the inverse neural network a weight. Before performing the learning process of the inverse neural network, providing a process of recalling, the process of recalling includes: multiplying the blood pressure characteristic values by weights, adding up, and adding an offset value to obtain a total value; Substituting the total value into an excitation function yields an output. Then, in the learning process, the output is subtracted from the known correct data to obtain a weight difference; and the weight is corrected according to the weight difference.

In an embodiment, the method further includes: finding, by the microcontroller, an AC waveform having a maximum amplitude from the AC waveforms; multiplying the maximum amplitude by a first multiple to obtain a first amplitude smaller than the maximum amplitude. And finding an AC waveform having a size closest to the first amplitude, and defining a first pressure value on the DC curve corresponding thereto as a systolic pressure; and multiplying the maximum amplitude by a second multiple to A second amplitude smaller than the maximum amplitude is obtained, and an AC waveform having the closest magnitude to the second amplitude is found, and a second pressure value on the DC curve corresponding thereto is defined as a diastolic pressure.

In an embodiment, the method further comprises: before calculating the diastolic pressure, the blood pressure measuring device first determines whether the number of the found feature values has reached a predetermined number; if the number of the feature values has reached a predetermined amount, calculating the diastolic pressure And a pulse rate; the systolic and diastolic pressures and pulse rate are transmitted to a mobile device.

The method for identifying an identity by using a blood pressure signal according to the present invention, by using a blood pressure measuring device itself or a computer or a mobile device, first extracts the characteristic value of the blood pressure signal measured by the blood pressure measuring device to identify the user identity, and then The feature values and identification results are transmitted to the cloud system, and cloud computing is performed to increase the recognition capability of the identity recognition algorithm. After the user measurement is completed, the measurement result can be stored in the user account of the local machine or the cloud system. Thus, the present invention can provide a convenient method for automatically storing blood pressure measurement data and user data.

The above and other technical contents, features and advantages of the present invention will be apparent from the following detailed description of the preferred embodiments. The directional terms mentioned in the following embodiments, such as upper, lower, left, right, front or rear, etc., are only used to refer to the directions of the accompanying drawings. Therefore, the directional terms are used for illustration only and are not intended to limit the invention.

FIG. 1 is a schematic flow chart of an embodiment of the present invention. The method for identifying an identity using the blood pressure signal of the present invention can be implemented by using a blood pressure measuring device, a mobile device and a cloud system to form a system architecture. First, the blood pressure signal is extracted from the blood pressure measuring device (S110), and the blood pressure characteristic value is calculated first (hereinafter, the blood pressure characteristic value is simply referred to as the characteristic value) (S111); then, it is first determined whether the blood pressure measuring device has the identification. Function (S120); if yes, the identification result is calculated (S130); if there is no identification function, it is judged whether the mobile device has the identification function (S121). Then, the mobile device calculates the identification result; if the mobile device still does not, it determines whether there is a connection to the network at present (S122); if the network is not connected, the feature value is put into a waiting state, and the connection is continued until the network is connected ( S123); if the network is connected, it is passed to the cloud system, and the feature value is taken as an input, and the cloud computing technology is used to execute the identity recognition algorithm, and the identity recognition result is calculated (S124), and then the identification is displayed by the mobile device. Result (S140). Then, the user judges whether the identification result is correct (S141); if the recognition error occurs, the user manually selects the correct identity and returns the error information to the identity recognition algorithm (S142); if the identification is correct or the correct identity has been selected , the correct identity and feature values are ready to be uploaded (S143). In the end, it will first determine whether there is a connection to the network (S150); if the network is not connected, the correct identity and feature values will enter the waiting state, and wait until the network is connected (S151); if connected to the network, it will be correctly identified. And the feature value is transmitted to the cloud system (S160). After receiving the message, the cloud system will change the structure of the identity recognition algorithm by obtaining multiple identification information such as correct identity, feature value and error information, and send the identification result back to the blood pressure measurement device, and then the blood pressure measurement. The measuring device uploads the user data together with the measurement result to the database, or the cloud system waits for the measurement result to be uploaded, and then stores it in the database (S170).

One of the focuses of the present invention is the sampling and calculation process of the eigenvalues in the blood pressure signal. Please refer to the concept of the blood pressure algorithm shown in Figures 2 to 3. 2 shows a blood pressure signal formed during the deflation of the blood pressure measuring device, which is a waveform diagram of pressure versus time, which includes a DC signal DC and an AC signal AC, wherein the DC standard The bit signal DC forms a DC curve whose pressure value decreases with time, and the AC signal AC forms a complex AC waveform W ₁ , W ₂ , W ₃ , wherein each AC waveform W ₁ , W ₂ , W ₃ has a rising band W _U connection. A falling band W _D and corresponds to a pressure value in the DC curve. It should be noted that the AC waveforms W ₁ , W ₂ , W ₃ corresponding to different time points t1, t2, and t3 are not the same, for example, the modes of the rising band W _U and the falling band W _D and both The area ratios are all different.

In an embodiment, during a period in which the DC curve drops from a high level pressure to a low level pressure, for example, from 150 mmHg to 40 mmHg, the microcontroller in the blood pressure measuring device is based on the measured The DC level signal DC and the AC signal AC select a plurality of complete AC waveforms W ₁ , W ₂ , W ₃ and perform sampling analysis. In FIG. 2, the first AC waveform W is selected after _1, when the cuff pressure value corresponding to the AC waveform by W ₁ P ₁ drop a particular pressure difference △ P reaches a certain pressure value P _2, then select this certain pressure value P ₂ corresponding to the AC waveform W _2. Whereby when the specific cuff pressure value P ₂ and then a drop of the same differential pressure △ P reaches the pressure value P _3, and then select the value of the pressure P ₃ corresponding to the AC waveform W _3. Similarly, when the DC pulse is lowered by a specific pressure difference ΔP and the DC curve reaches a specific pressure value, the microcontroller selects an AC waveform corresponding to the specific pressure value among the AC waveforms.

As shown in FIG. 3, for each selected AC waveform W ₁ , W ₂ , W ₃ , the valleys of the AC waveforms W ₁ , W ₂ , W ₃ are taken as a starting point, and a sampling frequency is used for sampling. The pressure value per point of the rising band W _U of the AC waveforms W ₁ , W ₂ , W ₃ is added from the valley W _T to the peak W _C , that is, the rising band W _U is integrated to calculate the area (Sum1). Taking the peak W _C as the starting point, sampling with the sampling frequency of 50 Hz, the pressure value of each point of the falling band W _D is added to the trough of the waveform, that is, the area of the falling band W _D is integrated to calculate the area (Sum2). . In order to enlarge the difference of the signal and facilitate the identification of the identity, the area of the rising band W _U and the falling band W _D is calculated as an individual square, and the two eigenvalues, that is, the square area of the two areas (Sum1*Sum1) and (Sum2*) are obtained. Sum2).

As shown in FIG. 4, it is a schematic flowchart of establishing an identity recognition algorithm according to an embodiment of the present invention. The blood pressure measuring device uses the blood pressure algorithm shown in FIGS. 2 and 3 in advance to find a plurality of characteristic values by using the entire blood pressure signal and its waveform peak value. After the blood pressure signal is processed by the blood pressure algorithm to obtain the feature value, an identity recognition algorithm is then started using the mobile device or the cloud system. These eigenvalues are taken as the input values of the identity recognition algorithm, and after the operation, the identification output of the identity is obtained. The process of establishing the identity recognition algorithm is as follows: firstly, multiple feature values in the blood pressure signal are found (S210); using these feature values as the input value of the identity recognition algorithm (S220), the neural network is performed in the cloud system. The learning process calculates all the weights (S230); performs a neural network-like recall process in the mobile device, and uses the weights to calculate the identification result (S240); and then, the user determines whether the identification result is correct, and the identification result does not match. At the time (S250), the cloud system is returned (S260), the learning process is performed again, the previous weight is changed (S270), and the accuracy of the identification result is improved (S280).

FIG. 5 illustrates the flow of the identity recognition algorithm of the present invention by using a back-transition neural network in more detail, taking 24 eigenvalues obtained by selecting 12 AC waveforms as examples. First build a three-layer network, and the weight of each node forwarded (indicated by the arrow), the three layers are set to:

The first layer, 24 nodes (representing 24 inputs)

The second layer, 5 nodes (representing the hidden layer)

The third layer, 2 nodes (representing 2 outputs)

At the end of each layer, a node with a bias is placed, so there are [24 (input layer) +1 (Bias)] + [5 (hidden layer) +1 (Bias)] + [2 (output layer). ) +1 (Bias)] = 34 nodes, and, 25 (input layer + Bias) * 5 (hidden layer) + 6 (hidden layer + Bias) * 2 (output layer) = 137 weight values.

The recall process is a process of forward transfer, the steps are as follows:

In the first step, multiply all the input values by a certain weight, add up, and add a partial weight.

In the second step, the sum value is substituted into an excitation function, which is a function with a maximum value of 1, a minimum value of -1, and an input value of a non-linear value, so the output will be between +1 and -1. between.

These two steps, which occur from the next layer to the next layer, are not corrected to the weight value at all, so they are called the recall process.

The learning process is a reverse transfer process with the following steps:

In the first step, during training, the output calculated by the process is subtracted from the expected output value, and the gradient descent is used to find the weight of ownership or the rate of change of weight.

In the second step, the weight difference or the weight change rate is changed to the original weight value, so that the weight value is closer to the ideal value, thereby achieving the purpose of learning.

Then, two parameters are used in the convergence process, one is the learning speed, that is, the displacement coefficient of the gradient steep drop method; the other is the inertia term, that is, the coefficient of the oscillation phenomenon in the convergence process is improved.

Finally, this embodiment sets the identification between three people. Therefore, the output will be divided into three kinds of identification results with 0.5 as the middle boundary, representing the identity of three people:

(0, 0), decimal = 0

(0, 1), decimal = 1

(1, 0), decimal = 2

Summarize the aforementioned training process, as shown in Figure 6. A target output is written into an application in advance. The target output is data known to train a user, and the user's feature value is passed to the program for training. In the process of recalling, each feature value is multiplied by a certain weight, and then added to an offset value to obtain a total value, and then the total value is substituted into an excitation function to calculate an output result. . In the following learning process, the output result is subtracted from the target output, and the subtracted value is substituted into the gradient steep drop method to find the weight difference or the weight change rate, and the weight difference or the weight change rate is used to correct the recall process. The original weight in . In this repeated training, when the number of trainings is sufficient, the weight will be close to the ideal value, which means that the application is a trained identity recognition algorithm. After that, as long as the feature value is input, after the result is calculated by the recall process, the user's identity can be recognized.

In this way, when a user measures the blood pressure by the blood pressure measuring device, a new identification result is added; the newly added identification result and a known correct data are uploaded to a cloud system; and in the cloud system, A new learning process and known correct data are used to perform a learning process of the inverse neural network to correct a weight of the inverse neural network.

FIG. 7 is a blood pressure measuring device 300 suitable for performing the processes described in FIG. 1 to FIG. 6. The hardware structure is divided into four parts: a microcontroller 320, a blood pressure signal capturing circuit 340, a peripheral control unit 360, and an antenna module. Group 380. The microcontroller 320 adopts a Bluetooth system single chip, which internally includes a core processor 322, a low power Bluetooth transmitting and receiving unit 324, an analog to digital converter (DAC) 326 (Analog to Digital Converter, ADC) and a universal input. /Output pin 328 (General Purpose Input/Output, GPIO). The blood pressure signal acquisition circuit 340 mainly amplifies and filters the signal of the pressure sensor 341, and then transmits it to the analog digital converter 326 of the microcontroller 320 for processing. The antenna module 380 is a low power Bluetooth transmitting and receiving unit 324 that uses a ceramic antenna 382 (antenna) and a balun 384 (balum) and is connected to the microcontroller 320. The peripheral control unit 360 includes a Bluetooth pairing button 362, a solenoid valve 364, a pneumatic pump 366 and a LED indicator 368 connected to the universal input/output pin 328.

The blood pressure signal acquisition circuit 340 has an analog filtering function. When the cuff 310 is deflated, the gas inside the cuff 310 is connected to the pressure sensor 341 in the blood pressure signal acquisition circuit 340, and the pressure sensor 341 converts the air pressure into a small voltage change. . The converted voltage signal is processed by a differential amplifier 342 (Differential Amplifier), and the amplified signal includes a DC level signal (DC) and a small AC signal (uAC), and the signal passes through a low pass filter 343 ( Low Pass Filter) Get a clean DC signal. The additional DC/AC mixed signal (DC+uAC) and the DC signal are subtracted from the DC signal by another differential amplifier 344, and the uAC signal is amplified, and then passed through another low pass filter 345 to obtain an AC signal (AC). The last DC signal and AC signal are then transmitted to the microcontroller 320 for processing.

In terms of the firmware architecture, the blood pressure measurement device 300 implements the blood pressure algorithm and the Bluetooth low energy transmission protocol using a single low power Bluetooth (Bluetooth Low Energy) system. For example, a layered stacked single-chip firmware architecture includes an application layer and a Generic Attribute Profile (GATT). The application layer provides a multiplex mode for various tasks (Tasks), and tasks can be set to trigger and operate at different cycle times. Each task is triggered with a different event (Event) to execute the code corresponding to each event. In one embodiment, blood pressure algorithms, pump and solenoid control, power management, and IO control can be written in the application layer, while the blood pressure transfer packet format is written in the common attribute setting layer. Moreover, the firmware structure of the blood pressure measuring device is divided into a main program and a plurality of trigger events.

As shown in FIG. 8, when the Bluetooth button on a mobile device is pressed, the mobile device notifies the blood pressure measuring device to start measurement by wireless transmission, and triggers an event (S400). During the pump inflation process, the microcontroller 320 will immediately process the DC and AC signals (S410), and then use the DC signal to determine whether the pump charge reaches an upper pressure value (Threshold) (S420), if the upper limit pressure value is reached. Then close the pump to stop inflation (S430). When the vent valve is deflated, the microcontroller 320 will process the DC and AC signals in real time (S440), and substitute the processed DC signal and the AC signal into the blood pressure algorithm to find the peak value and the eigenvalue in the AC signal. The current pressure value, characteristic value and measurement process data are transmitted to the mobile device end (S450) through the low-power tooth decay. The blood pressure algorithm determines whether a maximum peak-to-peak value (S460) is detected. If the largest peak-to-peak value is found, the systolic pressure (S461) can be calculated and the systolic pressure is transmitted. The low power tooth decay is transmitted to the mobile device terminal (S462). The blood pressure algorithm also detects whether a predetermined number of eigenvalues are completely found (S470), and if a predetermined number of eigenvalues are completely found, the diastolic blood pressure and the pulse rate (S471) are calculated, and the diastolic blood pressure and the pulse rate are low. The power consumption is transmitted to the mobile device (S472). Finally, the event trigger is turned off, the solenoid valve is turned on (S480), and the blood pressure measurement is ended (S490).

Regarding steps S461 and S471 for calculating the systolic blood pressure and the diastolic blood pressure, referring still to FIG. 2, when the amplitude of the pulse-pulsed zone is maximized, the pressure is closest to the Mean Arterial Pressure (MAP). Therefore, the blood pressure measuring device of the present invention uses its microcontroller to detect the largest peak-to-peak value from these AC waveforms, in other words, find an AC waveform having the largest amplitude to correspond to the average voltage on the DC curve. . Set the maximum amplitude to "1", multiply 1 by a multiple of X, and find the DC curve pressure value P _X corresponding to the AC waveform closest to the "X" amplitude on the DC curve. Pressure (S461). Similarly, after multiplying 1 by a multiple of Y, the AC waveform closest to the amplitude of the "Y" amplitude is searched for, and the corresponding DC curve pressure value P _{Y is} the diastolic pressure (S471).

As shown in Figure 9, the method of the present invention is implemented in a system architecture 500. The Blood Pressure Monitor (BPM) is a hardware device for measuring blood pressure. When measuring blood pressure, it records the waveform of the blood pressure signal and transmits the data via Bluetooth Low Energy (BLE). ) Transfer to Mobile Devices. The mobile device 540 can control the opening and closing of the blood pressure measuring device 300 through Bluetooth, and can also control the upper limit of the inflation of the blood pressure measuring device 300. After measuring the blood pressure and receiving the data, the Bluetooth is automatically disconnected, and Analyze which user measured the data, identify the user's identity, the mobile device 540 will display the measured data and identification results and store the data in the application and cloud system 560 database (Cloud Database )in. The database is mainly stored separately according to the user's data, which is convenient for the user to inquire about the measurement record.

10A to 10E show the flow of an application in a mobile device. In addition to the identity algorithm described above, the application also includes a mechanism for creating new members. FIG. 10A is an account creation screen 600 provided by the mobile device for establishing a plurality of account numbers. It is worth mentioning that in the mechanism for creating new members, each account password is based on one family; in other words, each account corresponds to multiple members in the same family. FIG. 10B is a screen 610 of creating a new member provided by the mobile device for establishing information of each member of the family and storing it in the same account. If you have previously created an account password in your mobile device, you can log in directly without having to enter it again. If it is used for the first time and has not previously established an account, you can enter the account and password first, and do the registration (Register) action.

Figure 10C shows a screen 620 of the measurement usage process. At the time of the equivalent measurement, the mobile device 540 is connected to the blood pressure measuring device 300 by pressing a Bluetooth button 621. When the connection is successful, the power button 622 transparency becomes 1, indicating a "capable of use" state, at which time the power button 622 can be pressed to activate the blood pressure measuring device 300. During the measurement process, the love icon 623 will be zoomed in and out according to the heart beat, and will also be accompanied by a heartbeat sound, so that the user can feel the beat of his heart, and the middle state bar 624 will be pressed according to the pressure. The degree of completion of the pressure detection measurement of the pulse zone 310. When the equivalent measurement is completed, the calculation result of the measurement and the identification result are displayed.

Figure 10D shows the generation of an identified member from among these members after the measurement is completed and the identity recognition algorithm is executed. At this point, the application provides a selection screen 630 that includes the identified members and other members of the family. For example, the selection screen 630 of FIG. 10D has three buttons with user photos. Moreover, the recognized member A is represented by a pattern having different transparency, for example, the transparency of the photo 631 of the recognized member is adjusted to 1, and the transparency of the photos 632 and 633 of other users is adjusted to 0.3. When the user determines that the identified member is incorrect, the selection screen 630 provides the user with a correct member to select from. Therefore, the user can select his or her identity, and when the selection is completed, a successful archiving screen will appear above.

11A and 11B show that the data in the download has two manifestations. FIG. 11A shows the history record in a table view 640 (tableview), including systolic blood pressure, diastolic blood pressure, heart beat, date, user name and use. Avatar. The format of each cell is the same, and the settings will update all the data every two seconds. Fig. 11B is a dynamic line graph 650 with a lot of animation functions, which are divided into three types of data to display, systolic blood pressure, diastolic blood pressure, and heart beat. Also, a button selection method is provided to allow the user to select the information they want to see. The storage function is to upload data to the cloud database through the network, reduce the waste of the mobile device memory space, and make the program more efficient.

In addition to being able to operate on a local machine, the method of the present invention can also be combined with cloud computing. For example, the blood pressure measuring device extracts the characteristic value of the blood pressure signal, and then uses the blood pressure measuring device itself or a local device such as a computer or a mobile device to identify the user's identity, and the local device Transfer the feature values and recognition results to the cloud system. The cloud system performs cloud computing according to the received feature values and identification results to change the structure of the identity recognition algorithm. The cloud system can further send the found user account back to the blood pressure measuring device, and the blood pressure measuring device uploads the user account and the identification result to a database. Alternatively, the cloud system can directly wait for the measurement result to be uploaded, and store it together with the user account and the identification result in the database. This method of identifying the user's identity according to the blood pressure signal simplifies the previous login action applied to the blood pressure measuring device, which can facilitate the measurement of the blood pressure signal by each age group, and accurately record the user identity and its measurement. Value.

The advantages of the present invention are as follows: when the user wants to perform blood pressure measurement, it is not necessary to perform login and typing actions before each blood pressure measurement, and the errors caused by human operation are reduced, for example, human login and data recording, etc. , reducing the cost of manpower. The method of the invention can automatically log in and reduce the time for storing user data, and improve the convenience of the use of the blood pressure measuring device, in particular, the desire of the elderly to use. For example, after the user puts on the blood pressure measuring device, the blood pressure measuring device automatically recognizes the user's identity through the user's blood pressure signal characteristic value when the button for starting the measurement is pressed, and the measurement is completed. After the automatic measurement of the measured value, as the operation of the general sphygmomanometer, without the need to increase the hardware cost.

The above is only the preferred embodiment of the present invention, and the scope of the invention is not limited thereto, that is, the simple equivalent changes and modifications made by the scope of the invention and the description of the invention are All remain within the scope of the invention patent. In addition, any of the objects or advantages or features of the present invention are not required to be achieved by any embodiment or application of the invention. In addition, the abstract sections and headings are only used to assist in the search of patent documents and are not intended to limit the scope of the invention.

DC‧‧‧DC level signal
AC‧‧‧ exchange signal
W ₁ , W ₂ , W ₃ ‧‧‧ AC waveform
t ₁ , t ₂ , t ₃ ‧‧‧
P ₁ , P ₂ , P ₃ ‧‧‧pressure value △P‧‧‧specific pressure difference
1‧‧‧Maximum amplitude
X, Y‧‧‧ amplitude
W _U ‧‧‧ rising band
W _D ‧‧‧Descent band
W _T ‧‧‧ Valley
W _C ‧‧·Crest
Sum1, Sum2‧‧‧ area
300‧‧‧Blood pressure measuring device
310‧‧‧Curve belt
320‧‧‧Microcontroller
322‧‧‧ core processor
324‧‧‧Low-power Bluetooth transmitting and receiving unit
326‧‧‧ Analog Digital Converter
328‧‧‧General purpose input/output pins
340‧‧‧Blood signal acquisition circuit
341‧‧‧pressure sensor
342‧‧‧Differential Amplifier
343‧‧‧Low-pass filter
344‧‧‧Differential Amplifier
345‧‧‧Low-pass filter
360‧‧‧ Peripheral Control Unit
362‧‧‧Bluetooth pairing button
364‧‧‧Solenoid valve
366‧‧‧Pneumatic pump
368‧‧‧Lighting diode indicator
380‧‧‧Antenna Module
500‧‧‧System Architecture
540‧‧‧Mobile devices
560‧‧‧Cloud System
600‧‧‧ account creation screen
610‧‧‧Create a picture of a new member
620‧‧‧Measure the picture of the use process
621‧‧‧Bluetooth button
622‧‧‧Power button
623‧‧‧Love icon
630‧‧‧Select screen
631‧‧‧Photos of identified members
632, 633‧‧‧Photos of other users
640‧‧‧Table view
650‧‧‧Dynamic line chart

FIG. 1 is a schematic diagram of an overall flow of a method for identifying an identity using a blood pressure signal according to an embodiment of the present invention.

2 to 3 are schematic views of a blood pressure algorithm according to an embodiment of the present invention.

FIG. 4 is a schematic diagram of a process of establishing an identity recognition algorithm according to an embodiment of the present invention.

FIG. 5 is a schematic diagram of an inverted transit neural network architecture used to establish an identity recognition algorithm according to an embodiment of the present invention.

FIG. 6 is a schematic diagram of a training process for establishing an identity recognition algorithm by using a reverse neural network according to an embodiment of the present invention.

FIG. 7 is a schematic diagram of a hardware structure of a blood pressure measuring device according to an embodiment of the present invention.

FIG. 8 is a schematic flow chart of a firmware of a blood pressure measuring device according to an embodiment of the present invention.

FIG. 9 is a schematic diagram of a system architecture according to an embodiment of the present invention.

10A to 10E are schematic diagrams showing the flow of a software program in a mobile device according to an embodiment of the present invention.

11A and 11B are schematic diagrams showing the manner of representation of measurement data according to an embodiment of the present invention.

no

no

DC‧‧‧DC level signal

AC‧‧‧ exchange signal

W ₁ , W ₂ , W ₃ ‧‧‧ AC waveform

_{t 1, t 2, t 3} ‧‧‧ time point

P ₁ , P ₂ , P ₃ ‧‧‧ pressure values

△P‧‧‧specific pressure difference

1‧‧‧Maximum amplitude

X, Y‧‧‧ amplitude

Claims (10)

A method for identifying an identity by using a blood pressure signal is applied to a blood pressure measuring device having a venous band, a pressure sensor and a microcontroller, comprising: the pressure sensor detecting the The pressure band of the cuff is changed during the deflation process to form a blood pressure signal, which includes a constant current signal and an alternating current signal, wherein the DC level signal forms a DC curve whose pressure value decreases with time. The signal forms a complex AC waveform, wherein each of the AC waveforms has a rising band connected to a falling band and corresponds to a pressure value in the DC curve; the DC curve is reached each time the pressure band drops by a specific pressure difference a specific pressure value, the microcontroller selects an AC waveform corresponding to the specific pressure value among the AC waveforms; for the selected AC waveform, the microcontroller calculates the area of the rising band and the Decreasing the area of the band to define a complex blood pressure eigenvalue; establishing an identity recognition algorithm with a reverse neural network; and Characterized in that the blood pressure value entered identity identification algorithm, and outputs a recognition result. The method for identifying an identity by using a blood pressure signal according to claim 1, wherein the step of selecting the AC waveform by the microcontroller is during a period in which the DC curve drops from a high level pressure to a low level pressure. In progress. The method for identifying an identity by using a blood pressure signal as described in claim 1, further comprising: providing a software program, including the identity recognition algorithm and a mechanism for creating a new member, wherein the mechanism for creating a new member includes: providing An account creation screen for establishing a plurality of accounts, each of which corresponds to a plurality of members in the same family; providing a screen for creating a new member for establishing information of each member of the family and storing in the account And, after performing the identity algorithm, generate an identified member from among the members. The method for identifying an identity by using a blood pressure signal as described in claim 3, further comprising: after performing the identity recognition algorithm, providing a selection screen, the selection screen including the identified member and other ones in the family a member, wherein the identified member is represented by a pattern having different transparency; and when the user determines that the recognized member is incorrect, the selection screen provides the user with a correct member selected therefrom. The method for identifying an identity by using a blood pressure signal according to claim 1 of the patent application, further comprising: sampling at a sampling frequency in the rising band and the falling band of the selected AC waveform. The method for identifying an identity by using a blood pressure signal as described in claim 5, further comprising: calculating an area of the rising band and an area of the falling band to obtain a square of two areas, each of the areas. The square value is one of these blood pressure characteristic values. The method for identifying an identity by using a blood pressure signal according to claim 1, wherein the step of establishing the identity recognition algorithm by using the inverse neural network comprises: measuring a blood pressure measuring device by a user each time After the blood pressure, a new identification result is added; the newly added identification result and a known correct data are uploaded to a cloud system; and in the cloud system, the newly added identification result and the known correctness are utilized. The data is subjected to a learning process of the inverse neural network to correct a weight of the inverted neural network. The method for identifying an identity by using a blood pressure signal as described in claim 6 of the patent application, further comprising: providing a recalling process before performing the learning process of the inverse neural network, the recalling process comprising: After multiplying the weights respectively, adding up, adding an offset value, obtaining a total value; and substituting the total value into an excitation function to obtain an output result; in the learning process, The output is subtracted from the known correct data to obtain a weight difference; and the weight is corrected based on the weight difference. The method for identifying an identity by using a blood pressure signal according to claim 1 of the patent application, further comprising: finding, by the microcontroller, an AC waveform having a maximum amplitude from the AC waveforms; multiplying the maximum amplitude by a first a multiple to obtain a first amplitude smaller than the maximum amplitude, and to find an AC waveform having the closest magnitude to the first amplitude, and define a first pressure value on the DC curve corresponding thereto as one Squeezing pressure; and multiplying the maximum amplitude by a second multiple to obtain a second amplitude that is less than the maximum amplitude, and finding an AC waveform having the closest magnitude to the second amplitude, and corresponding thereto A second pressure value on the DC curve is defined as a diastolic pressure. The method for identifying an identity by using a blood pressure signal according to claim 9 of the patent application, further comprising: before calculating the diastolic pressure, the blood pressure measuring device first determining whether the number of the characteristic values found has reached a predetermined amount; When the number of the feature values has reached the predetermined amount, the diastolic pressure and a pulse rate are calculated; and the systolic pressure and the diastolic pressure and the pulse rate are transmitted to a mobile device. .