Blood pressure meter, personal authentication method in blood pressure meter, and program

Abstract

A sphygmomanometer (1) of the present application includes a cuff for pressing a measurement site of a subject, and measures blood pressure by observing the pressure of the cuff. The sphygmomanometer (1) includes a feature acquisition unit (220) that acquires feature information about a pressure change pattern that increases in rate and rises over time from the start of pressurization, during pressurization of the cuff, for a subject to be authenticated, and an authentication unit (240) that compares the acquired feature information with registration feature information about a user registered in advance, and performs individual authentication of the subject.

Description

Technical Field

[0001] This invention relates to blood pressure monitors, and more specifically, to blood pressure monitors with personal authentication functionality. Furthermore, this invention relates to a personal authentication method in a blood pressure monitor. Additionally, this invention relates to a program for causing a computer to execute such a personal authentication method in a blood pressure monitor. Background Technology

[0002] When a person uses a blood pressure monitor to measure their blood pressure, it is desirable to perform personal authentication (to confirm that the person being tested is the logged-in user) in order to establish a correspondence between the person being tested and the measured blood pressure value. Conventionally, methods for personal authentication via a blood pressure monitor include, for example, those disclosed in Patent Document 1 (Japanese Patent Application Publication No. 2010-110380) and Patent Document 2 (Japanese Patent Application Publication No. Hei 6-142065), which use a pattern of Korotkoff tone level changes. Furthermore, as disclosed in Patent Document 3 (Japanese Patent Application Publication No. 2003-299624), a remote diagnostic assistance system is known that extracts feature parameters from an electrocardiogram signal sent by a user to a server, performs personal authentication using a neural network classification method, and then performs a diagnosis.

[0003] Existing technical documents

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent Application Publication No. 2010-110380

[0006] Patent Document 2: Japanese Patent Application Publication No. 6-142065

[0007] Patent Document 3: Japanese Patent Application Publication No. 2003-299624 Summary of the Invention

[0008] The problem that the invention aims to solve

[0009] However, due to the significant deviations in each measurement of Korotkoff sounds and electrocardiogram (ECG) signals, accurate authentication is difficult. Furthermore, recently, blood pressure monitors that measure cuff pressure (internal pressure), such as those using oscillometric methods, have become widespread. In these cuff-pressure-based blood pressure monitors, there is a problem that when performing personal authentication using Korotkoff sounds and ECG signals, unnecessary components not used during pressure measurement (e.g., microphones and other sound detection devices, ECG electrodes, etc.) must be installed.

[0010] Therefore, the objective of this invention is to provide a blood pressure monitor that measures blood pressure by observing cuff pressure, capable of personal authentication with a simple structure and high accuracy. Furthermore, the objective of this invention is to provide a personal authentication method for such a blood pressure monitor. Additionally, a program is provided for enabling a computer to execute such a personal authentication method.

[0011] Problem-solving methods

[0012] To solve the aforementioned problem, the blood pressure monitor disclosed herein includes a cuff for compressing the measurement site of the subject, and blood pressure is measured by observing the pressure of the cuff, characterized in that...

[0013] have:

[0014] The pressure control unit controls the supply of fluid to the cuff to increase pressure or the discharge of fluid from the cuff to decrease pressure.

[0015] The pressure detection unit detects the pressure of the cuff;

[0016] The blood pressure calculation unit calculates blood pressure based on the output of the pressure detection unit;

[0017] The feature acquisition unit acquires feature information about the pattern of pressure change of the cuff over time from the start of pressure application, for the person being tested who is being authenticated;

[0018] The authentication department compares the acquired feature information with the login feature information of pre-logged-in users to perform personal authentication on the person being tested.

[0019] In this specification, "cuff pressure" refers to the pressure within the cuff (typically a fluid pocket embedded in the cuff). Hereinafter, "cuff pressure" will be simply referred to as "cuff pressure".

[0020] "Pressure variation pattern" refers to the pattern of pressure change in the cuff over time from the start of pressurization. In addition to a smooth, monotonically increasing component (DC component), the pressure in the cuff also includes a pressure variation component caused by the pulse wave at the measurement site.

[0021] "Performing personal authentication" refers to determining whether the person being authenticated is a pre-registered user (the person themselves). In this specification, the determination of whether the person being authenticated is a pre-registered user is based on whether the characteristic information of the pressure change pattern of the person being authenticated matches the login characteristic information of a pre-registered user. Furthermore, a "pre-registered user" can be a single person or multiple people.

[0022] "Login feature information" is typically feature information that is acquired in advance by the feature acquisition unit and stored in the storage unit before the personal authentication is performed.

[0023] In the blood pressure monitor disclosed herein, blood pressure measurement is performed as follows: With the cuff worn on the measurement site, a pressure control unit supplies fluid to the cuff to increase pressure or discharges fluid from the cuff to decrease pressure. During the pressurization process based on this pressurization or the subsequent depressurization process, a pressure detection unit detects the pressure of the cuff. A blood pressure calculation unit calculates the blood pressure based on the output of the pressure detection unit. Thus, blood pressure measurement is performed.

[0024] Furthermore, personal authentication of the test subject is performed as follows. The feature acquisition unit acquires feature information about the pressure change pattern—specifically, the rate of increase in pressure rise over time—as the test subject is pressurized during the pressurization process. The authentication unit compares this acquired feature information with login feature information for pre-registered users to perform personal authentication of the test subject. Here, it is empirically known that the pressure change pattern (particularly a DC pressure change pattern corresponding to the DC component of the cuff pressure and showing an increasing rate of increase over time) depends on the test subject's physical characteristics, such as the circumference of the measured area (thick or thin arm) and body composition (muscular or fatty body). For example, in cases where the circumference of the measured area is large (i.e., thick arm) compared to cases where the circumference is small (i.e., thin arm), the increased internal volume of the cuff (fluid bag) leads to a tendency for a smaller rate of increase (slope) in the cuff pressure when fluid is supplied to the cuff at a constant flow rate per unit time during pressurization. Furthermore, when the measured body part is fatty, compared to a muscular body, the measured area tends to flatten more easily in the high-pressure region, resulting in a smaller rate of increase (slope) of cuff pressure in the high-pressure region. Moreover, these tendencies can combine depending on the combination of the circumference of the measured body part, body type, etc., of each individual. These physical characteristics exhibit relatively small variations over short periods for each individual. In other words, these physical characteristics differ from Korotkoff sounds and ECG signals, resulting in small deviations in each measurement. Therefore, in this blood pressure monitor, by using characteristic information about the pressure change pattern determined by the individual's physical characteristics for personal authentication, highly accurate personal authentication of the individual is possible. Furthermore, this eliminates the need for unnecessary components not used during pressure observation (e.g., microphones or other sound detection devices, ECG electrodes, etc.) when performing personal authentication for the individual. Therefore, this blood pressure monitor can be constructed with a simple structure and at low cost.

[0025] In one embodiment of the blood pressure monitor, the characteristic is that,

[0026] The authentication department creates a feature space where the different feature quantities contained in the feature information about the pressure change pattern are set as coordinate axes.

[0027] In the feature space, points corresponding to the login feature information are calculated for the pre-logged-in user, and an allowable range is set around the points corresponding to the login feature information to identify the detected person as the pre-logged-in user.

[0028] In the feature space, when a point corresponding to the feature information obtained for the detected person enters the allowable range, the detected person is determined to be the pre-logged-in user. On the other hand, when a point corresponding to the feature information obtained for the detected person does not enter the allowable range, the detected person is determined not to be the pre-logged-in user.

[0029] In this embodiment of the blood pressure monitor, by appropriately setting the allowable range, it is possible to further determine with high accuracy whether the person being tested (the person being tested as the authentication object) is the pre-registered user.

[0030] In one embodiment of the blood pressure monitor, the characteristic is that,

[0031] have:

[0032] The first pressure detection unit extracts the DC component from the pressure of the cuff during the pressurization process and detects the DC pressure change pattern; and

[0033] The second pressure detection unit extracts the pressure variation component caused by the pulse wave at the measured location from the pressure of the cuff during the pressurization process of the cuff or during the depressurization process after the pressurization process, and detects the waveform pattern of each beat.

[0034] The feature acquisition unit, in addition to acquiring the DC feature quantity of the DC pressure change pattern, also acquires the feature quantity of each beat of the waveform pattern, as the feature information.

[0035] The authentication department, in addition to comparing the acquired DC characteristic quantity with the pre-registered DC characteristic quantity, also compares the acquired one-shot characteristic quantity with the pre-registered one-shot characteristic quantity to perform personal authentication of the person being tested.

[0036] "Characteristic quantity of DC" refers to the characteristic quantity of the pressure change pattern of DC. In addition, "characteristic quantity of one beat" does not refer to the pressure change pattern of DC, but rather to the characteristic quantity of the waveform pattern of each beat.

[0037] In this embodiment of the blood pressure monitor, a first pressure detection unit extracts a DC component from the cuff pressure during the inflation process and detects a DC pressure change pattern. Conversely, a second pressure detection unit extracts a pressure change component caused by the pulse wave at the measurement site from the cuff pressure during the inflation process or during the deflation process, detecting a waveform pattern for each beat. The feature acquisition unit acquires not only the DC feature quantity related to the DC pressure change pattern but also the feature quantity for each beat of the waveform pattern, as the feature information. The authentication unit compares the acquired DC feature quantity with pre-registered DC feature quantities and also compares the acquired beat feature quantity with pre-registered beat feature quantities to perform personal authentication of the subject. Thus, in addition to the DC feature quantity, the beat feature quantity is used as a judgment criterion, enabling more accurate personal authentication of the subject.

[0038] In one embodiment of the blood pressure monitor, the characteristic is that,

[0039] Under a pre-set first condition, the authentication department uses only the DC characteristic quantity instead of the one-time characteristic quantity in order to perform personal authentication of the tested person.

[0040] Here, as a "pre-set first condition," for example, a condition could be that the reliability of the feature quantity of the pulse during personal authentication is significantly higher or lower than the pulse during login of the login feature information.

[0041] In this embodiment of the blood pressure monitor, the authentication unit, under a pre-set first condition, uses only the DC characteristic quantity instead of the beat characteristic quantity for personal authentication of the subject. Here, the first condition is, for example, that the pulse rate during personal authentication is significantly higher or lower than the pulse rate when the registration characteristic information is entered. According to this embodiment of the blood pressure monitor, in cases where the reliability of the beat characteristic quantity is low (i.e., cases where the use of the beat characteristic quantity should be excluded), only the DC characteristic quantity can be used. As a result, a decrease in the accuracy of the personal authentication can be prevented.

[0042] In one embodiment of the blood pressure monitor, the characteristic is that,

[0043] have:

[0044] Storage department; and

[0045] The feature login unit performs the following control in advance before the personal authentication: during the compression of the cuff, it acquires feature information about the pressure change pattern of the subject, establishes a correspondence between the acquired feature information and the subject, and stores it in the storage unit as the login feature information.

[0046] In this embodiment of the blood pressure monitor, the feature registration unit performs the following control in advance before performing personal authentication. First, during the inflation of the cuff, feature information about the pressure change pattern of the subject is acquired. Then, the acquired feature information is associated with the subject and stored in the memory as the registered feature information. As a result, during the personal authentication phase, the authentication unit uses the registered feature information stored in the memory as a comparison benchmark to perform personal authentication of the subject.

[0047] In one embodiment of the blood pressure monitor, the characteristic is that,

[0048] have:

[0049] Storage Department:

[0050] A neural network has an input layer containing multiple nodes, multiple intermediate layers, and an output layer.

[0051] The neural network learns by weighting the nodes, so that when any one of the feature vectors representing feature information about the pressure change patterns of multiple teacher users is input to the input layer, the output layer outputs output information representing the teacher user corresponding to the input feature information, via the multiple intermediate layers.

[0052] The blood pressure monitor has a feature login unit that performs the following pre-controls before the personal authentication process: During the inflation of the cuff, for each user group including the multiple teacher users and / or users other than the teacher users, feature information about the pressure change pattern of that user is acquired. The acquired feature information is then input as a feature vector into the input layer of the neural network, so that the information presented in one of the multiple intermediate layers is associated with the user and stored in the storage unit as the login feature information.

[0053] During the personal authentication phase, the feature acquisition unit acquires feature information about the pressure change pattern of the subject being authenticated during the compression of the cuff.

[0054] The authentication unit inputs the acquired feature information as a feature vector into the input layer of the neural network, compares the information of a certain intermediate layer presented in the plurality of intermediate layers with the login feature information of each user in the user group logged in to the storage unit, and determines whether the detected person is a user included in the user group.

[0055] "Teacher user" refers to a user who provides feature information about the pressure change pattern for the learning of the neural network. "User group" includes multiple teacher users and / or users other than the teacher users.

[0056] In this embodiment of the blood pressure monitor, the feature registration unit performs the following control in advance before performing the personal authentication: First, during the inflation of the cuff, for each user group including the plurality of teacher users and / or users other than the teacher users, feature information about the pressure change pattern of that user is acquired. Next, for each user, the acquired feature information of that user is input as a feature vector into the input layer of the neural network, and the information presented in one of the plurality of intermediate layers is associated with the user and registered in the storage unit as the registration feature information. Thus, for each user, the registration feature information is registered in the storage unit.

[0057] During the personal authentication phase, the feature acquisition unit acquires feature information about the pressure change pattern of the subject being authenticated during the compression of the cuff. The authentication unit inputs this acquired feature information as a feature vector into the input layer of the neural network. Furthermore, the authentication unit compares the information presented in one of the multiple intermediate layers with the login feature information of each user in the user group logged into the storage unit to determine whether the subject being authenticated is a user within the user group. This allows for the authentication of the fact that the subject being authenticated is a user within the user group.

[0058] In one embodiment of the blood pressure monitor, the characteristic is that,

[0059] have:

[0060] When the updating unit determines that the detected user is a pre-logged-in user, it uses the feature information obtained for the detected user and updates the user's login feature information under a pre-set second condition.

[0061] As a "pre-set second condition," for example, one could be a condition that the login characteristic information should be updated if a certain period of time has elapsed since the login (or the last update) regarding the user's login characteristic information (e.g., a year, a period during which the characteristic information regarding the pressure change pattern may change significantly).

[0062] In this embodiment of the blood pressure monitor, when the updating unit determines that the person being tested is a pre-registered user, it updates the user's registration feature information using the feature information obtained for the person being tested, under a preset second condition. Here, the second condition is, for example, that a certain period has elapsed since the user's registration (or the last update) regarding the registration feature information (e.g., a period of six months during which the feature information regarding the pressure change pattern may change significantly). According to this embodiment of the blood pressure monitor, when the registration feature information should be updated, it can automatically update the registration feature information and maintain it in an appropriate state. As a result, it is possible to prevent a decrease in the accuracy of personal authentication.

[0063] In one embodiment of the blood pressure monitor, the characteristic is that,

[0064] The authentication department suspends the personal authentication under a pre-set third condition.

[0065] As a "pre-set third condition," for example, the condition that the detection of irregular pulse waves, etc., is considered to be a condition with low reliability of characteristic information about the pressure change pattern (characteristic quantities of the DC and / or characteristic quantities of the beat).

[0066] In this embodiment of the blood pressure monitor, the authentication unit suspends personal authentication under a pre-set third condition. Therefore, the third condition is, for example, the detection of an irregular pulse wave. According to this embodiment of the blood pressure monitor, personal authentication can be suspended when the reliability of characteristic information regarding such pressure change patterns (the characteristic quantity of the DC pulse and / or the characteristic quantity of a single beat) is low. As a result, a decrease in the accuracy of the personal authentication can be prevented.

[0067] In another aspect, in the personal authentication method of the blood pressure monitor disclosed herein, the blood pressure monitor comprises a cuff for compressing the measurement site of the subject, and blood pressure is measured by observing the pressure of the cuff, characterized in that...

[0068] The blood pressure monitor has the following features:

[0069] The pressure control unit controls the supply of fluid to the cuff to increase pressure or the discharge of fluid from the cuff to decrease pressure.

[0070] The pressure detection unit detects the pressure of the cuff; and

[0071] The blood pressure calculation unit calculates blood pressure based on the output of the pressure detection unit.

[0072] In the aforementioned personal authentication method,

[0073] For the test subject being certified, during the compression process of the cuff, characteristic information about the pattern of cuff pressure change over time from the start of compression is acquired.

[0074] The acquired feature information is compared with the login feature information of the pre-logged-in user to perform personal authentication on the detected person.

[0075] According to the personal authentication method in the blood pressure monitor disclosed herein, personal authentication can be performed with high accuracy in a blood pressure monitor with a simple structure.

[0076] In another aspect, the program disclosed herein is a program for causing a computer to perform a personal authentication method in the blood pressure monitor.

[0077] By having a computer execute the program disclosed herein, the personal authentication method in the blood pressure monitor can be implemented.

[0078] Invention Effects

[0079] As can be seen from the above, the blood pressure monitor according to this disclosure can perform personal authentication with high accuracy using a simple structure. Furthermore, the personal authentication method in the blood pressure monitor according to this disclosure can perform personal authentication with high accuracy in a blood pressure monitor with a simple structure. Additionally, the personal authentication method in the blood pressure monitor can be implemented by having a computer execute the program of this disclosure. Attached Figure Description

[0080] Figure 1 This is a block diagram illustrating the structure of a blood pressure monitor according to one embodiment of the present invention.

[0081] Figure 2 This is a block diagram illustrating the functional structure of the CPU in the aforementioned blood pressure monitor.

[0082] Figure 3 This is a diagram illustrating the blood pressure measurement process of the aforementioned sphygmomanometer.

[0083] Figure 4 This is a diagram illustrating the process of logging into the blood pressure monitor described above.

[0084] Figure 5 This is a diagram illustrating the processing flow of the above-mentioned blood pressure monitor measurement mode.

[0085] Figure 6 (B) is a graph showing the change in cuff pressure over time from the start of pressurization. Figure 6 (A) is a graph representing the pressure variation components (pulse wave signal) of the cuff pressure obtained over time from the start of pressurization.

[0086] Figure 7 (A) is a graph showing the change in DC pressure over time from the start of pressure application for a user with a muscular physique and thick arms. Figure 7 (B) is a graph showing the DC pressure change over time from the start of pressure application for a user with muscular build and thin arms.

[0087] Figure 8 (A) is a graph showing the change in DC pressure over time from the start of pressurization for a user with a fatty body type and thick arms. Figure 8 (B) is a graph showing the DC pressure change over time from the start of pressurization for a user with a fatty body type and thin arms.

[0088] Figure 9 It is a graph representing the permissible range (the range acceptable to the user) in the feature space used to determine whether the person being tested as the authentication object is a logged-in user.

[0089] Figure 10 It is a diagram illustrating the characteristic information (characteristic quantity of one beat) of the waveform pattern for each beat.

[0090] Figure 11 (B) is a diagram showing the structure of the neural network of the modified blood pressure monitor after the above-mentioned blood pressure monitor has been modified. Figure 11 (A) Figure 11 (C) is a diagram representing the input signal (feature vector) and teacher signal provided to the neural network during the learning phase, respectively.

[0091] Figure 12 It is a diagram used to illustrate the feature information (feature vector) of the DC pressure change pattern provided to the above neural network. Detailed Implementation

[0092] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0093] (Structure of a blood pressure monitor)

[0094] Figure 1 The appearance of a blood pressure monitor 1 according to one embodiment of the present invention is shown. The blood pressure monitor 1 generally includes: a blood pressure measuring cuff 20, which is worn by wrapping around the bar-shaped measuring part (e.g., the upper arm) of the subject; and a body 10, which is equipped with elements for blood pressure measurement.

[0095] The aforementioned cuff 20 is a general cuff, consisting of a fluid bag 22 sandwiched between a thin strip of outer fabric 21 and inner fabric 23, and is constructed by sewing or welding the periphery of these outer fabrics 21 and inner fabrics 23.

[0096] The main body 10 is equipped with a CPU (Central Processing Unit) 100 as a processor, a display 50, an operation unit 52, a memory 51 as a storage unit, a power supply unit 53, a pressure sensor 31, a first filter unit 311 and a second filter unit 315, a pump 32, a pump drive circuit 320, a valve 33, and a valve drive circuit 330. In this example, the air pipe 39a connected to the pressure sensor 31, the air pipe 39b connected to the pump 32, and the air pipe 39c connected to the valve 33 merge into a single air pipe 39, which is connected to the fluid bag 22 inside the cuff 20 in a fluid-flowable manner. Hereinafter, the air pipes 39a, 39b, and 39c mentioned above will be collectively referred to as air pipe 39.

[0097] In this example, the display 50 is composed of an LCD (Liquid Crystal Display) and displays prescribed information based on control signals from the CPU 100. In this example, the display 50 displays systolic blood pressure (SBP, mmHg), diastolic blood pressure (DBP, mmHg), pulse rate (beats / minute), and also displays the results of the individual's authentication. Furthermore, the display 50 may be composed of an organic EL (Electroluminescence) display or may include LEDs (Light Emitting Diodes).

[0098] In this example, the operation unit 52 includes a power switch 52A for turning the power on or off the blood pressure monitor 1, a measurement switch 52B for accepting an indication of the start of the blood pressure measurement mode, and a login switch 52C for accepting an indication of the start of the login mode. Operation signals corresponding to the user's instructions are input to the CPU 100. Specifically, when the power switch 52A is turned on, the blood pressure monitor 1 is in a power-on state, capable of accepting user operations on the measurement switch 52B and the login switch 52C. When the measurement switch 52B is pressed in the power-on state, the blood pressure monitor 1 performs the measurement mode processing (including blood pressure measurement and personal authentication processing) described later. When the login switch 52C is pressed in the power-on state, the blood pressure monitor 1 performs the login mode processing (performing user-related login processing) described later. In the following description, the blood pressure monitor 1 is set to the power-on state.

[0099] The memory 51 stores data for the program used to control the blood pressure monitor 1, setting data for setting various functions of the blood pressure monitor 1, data for blood pressure measurement results, and registration feature information as a comparison benchmark for personal authentication. Additionally, the memory 51 is also used as working memory when the program is executed.

[0100] The CPU 100 controls the overall operation of the blood pressure monitor 1 according to the program stored in the memory 51. Specifically, as... Figure 2 As shown, the CPU 100 functions as the first pressure detection unit 131, the second pressure detection unit 132, the pressure control unit 133, the blood pressure calculation unit 134, and the display processing unit 135. Furthermore, the CPU 100 functions as the input unit 210, the feature acquisition unit 220, the feature registration unit 230, the authentication unit 240, and the update unit 250 included in the authentication function module 200. The specific functions of each unit will be explained later.

[0101] In this example, Figure 1 The pressure sensor 31 shown includes a piezoelectric resistive semiconductor pressure sensor and an oscillation circuit. The aforementioned piezoelectric resistive semiconductor pressure sensor transmits the pressure (cuff pressure) Pc (refer to) within the fluid bag 22 of the cuff 20 via air piping 39. Figure 6 The (B) output is a resistor based on the piezoelectric effect. The above-mentioned oscillation circuit oscillates at an oscillation frequency corresponding to the resistance from the above-mentioned semiconductor pressure sensor, and outputs a frequency signal containing this oscillation frequency as a signal representing the cuff pressure Pc. The first filter unit 311 includes a low-pass filter (LPF) 312 and an A / D converter 314, which extracts the DC component (referred to as "DC component Pdc") from the frequency signal representing the cuff pressure Pc and converts it into a digital value. The second filter unit 315 includes a high-pass filter (HPF) 316 and an A / D converter 318, which extracts the pressure variation component (pulse wave signal) Pm (reference) caused by the pulse wave shown at the measured site from the frequency signal representing the cuff pressure Pc. Figure 6 The DC component Pdc and pulse wave signal Pm are converted into digital values. The digitized DC component Pdc and pulse wave signal Pm are input to the CPU 100. The CPU 100 functions as the first pressure detection unit 131, detecting the DC pressure change pattern (e.g., based on the input DC component Pdc) Figure 7 The pressure change pattern PdcU1 shown in (A) is also included. Furthermore, the CPU 100 functions as a second pressure detection unit 132, detecting the waveform pattern of each pulse based on the input pulse wave signal Pm (e.g., ...). Figure 6The waveform pattern Pm1 for each beat shown in (A). The pressure sensor 31, the first filter section 311, the second filter section 315, the first pressure detection section 131, and the second pressure detection section 132 together constitute the pressure detection section for detecting the pressure of the cuff 20.

[0102] Figure 1 The pump 32 shown is driven by the pump drive circuit 320 based on a control signal provided from the CPU 100 (which functions as a pressure control unit 133), and supplies air to the fluid bag 22 built into the cuff 20 through the air pipe 39. As a result, the pressure of the fluid bag 22 (cuff pressure Pc) is increased.

[0103] Valve 33 is a normally open solenoid valve that is driven by valve drive circuit 330 based on control signals provided by CPU 100 (which functions as pressure control unit 133) to control the cuff pressure by discharging or sealing air into fluid bag 22 through air pipe 39.

[0104] The power supply unit 53 supplies power to the CPU 100, display 50, memory 51, pressure sensor 31, pump 32, valve 33, and other components within the main body 10.

[0105] (Blood pressure measurement)

[0106] Figure 3 The processing of the login mode described later is shown. Figure 4 ), and the processing of the measurement mode described later ( Figure 5 The procedures for blood pressure measurement are performed separately in each of the following steps.

[0107] When the cuff 20 is worn on the area being measured, the user (subject) indicates the start of the measurement mode by using the measurement switch 52B located on the main body 10. Figure 3 In step S11), the CPU 100 first performs initialization (step S12). Specifically, the CPU 100 initializes the processing memory area and, with the pump 32 stopped and the valve 33 open, adjusts the pressure sensor 31 to 0 mmHg (setting the atmospheric pressure to 0 mmHg).

[0108] Next, the CPU 100 functions as the pressure control unit 133, closing valve 33 (step S13), driving pump 32, and initiating pressurization of cuff 20 (step S14). That is, the CPU 100 supplies air from pump 32 through air piping 39 to the fluid bag 22 embedded in cuff 20 at a certain flow rate per unit time. Simultaneously, pressure sensor 31 detects the pressure (cuff pressure) Pc within cuff 20 (fluid bag 22) through air piping 39. Here, as... Figure 6 (A) and Figure 6As shown in (B), in the cuff pressure Pc detected by the pressure sensor 31, in addition to the component that increases smoothly and monotonically (DC component Pdc), there is also a variable component (pulse wave signal) Pm caused by the pulse wave. The CPU 100 functions as the first pressure detection unit 131, and detects the DC pressure change pattern (e.g., the DC component Pdc input through the first filter unit 311) that increases in rate of rise and increases (i.e., the slope gradually increases and curves downward in a convex shape) over time from the start of pressurization, based on the DC component Pdc. Figure 7 The pressure change pattern PdcU1 shown in (A) Figure 3 (Step S15). Along with this, the CPU 100 controls the pressurization rate of the pump 32 based on the output of the first pressure detection unit 131. This pressurization compresses and blocks blood flow to the artery at the measured site.

[0109] Next, based on the output of the first pressure detection unit 131, the CPU100 determines when the cuff pressure Pc reaches a preset value Pu (in this example, such as...). Figure 7 When the pump 32 is stopped (as shown in (B) with Pu = 200 mmHg, which is sufficiently higher than the assumed blood pressure value of the subject), the pump is stopped. Figure 3 Step S16).

[0110] Next, the CPU100, acting as the pressure control unit 133, slowly opens the valve 33. This reduces the pressure on the cuff Pc at a roughly constant rate. Figure 3 Step S17). During this decompression process, the CPU 100 functions as the second pressure detection unit 132, detecting the pulse wave signal Pm (containing the waveform pattern Pm1 for each beat) input through the second filter unit 315. Figure 3 Step S18). Then, the CPU 100 functions as the blood pressure calculation unit 134, and attempts to calculate the blood pressure values ​​(systolic blood pressure SBP and diastolic blood pressure DBP) based on the pulse wave signal Pm acquired at that time point, for example, using a known oscillometric method. Figure 3 (Step S19). In addition, in this example, the CPU100 calculates the pulse rate [beats / minute] based on the pulse wave signal described above.

[0111] The CPU100 is unable to calculate blood pressure and pulse rate due to insufficient data. Figure 3 If step S20 is "No", repeat steps S17 to S20 until calculation is possible.

[0112] Thus, once the blood pressure and pulse rate can be calculated (step S20 is "yes"), the CPU100 functions as the pressure control unit 133, opening the valve 33 to control the rapid expulsion of air from the cuff 20 (fluid bag 22) (step S21).

[0113] Then, the CPU 100 controls the storage of blood pressure and pulse count in memory 51 (step S22).

[0114] In addition, in the example above, the blood pressure and pulse rate were calculated during the decompression of the cuff 20 (fluid bag 22), but it is not limited to this; the blood pressure and pulse rate can also be calculated during the inflatation of the cuff 20 (fluid bag 22).

[0115] (Handling login modes)

[0116] Figure 4 The process of the login mode is shown before the personal authentication of the blood pressure monitor 1. Here, a new user who has not yet registered a user login number is in the wearing state as the subject of the test, with the cuff 20 worn on the upper arm as the measurement site. In this state, when the user indicates the start of the login mode by means of the login switch 52C set on the main body 10, the CPU 100 performs the login mode processing as follows.

[0117] First of all, Figure 4 In step S101, the CPU 100 causes the display 50 to show a screen for the user to select a new user login number. For example, the display 50 shows a candidate user login number such as "Set the new user login number as U1?". At this time, when the user briefly presses (within 1 second) the login switch 52C, "U1" is registered as the user's user login number in the memory 51. Instead, when the user presses (for more than 3 seconds) the login switch 52C, the CPU 100 causes the display 50 to show another candidate user login number such as "Set the new user login number as U2?". At this time, when the user briefly presses the login switch 52C, "U2" is registered as the user's user login number in the memory 51. In this way, the CPU 100 causes the display 50 to sequentially display the candidate user login numbers U1, U2, ..., and registers the user's original user login number in the memory 51 according to the user's selection. Furthermore, for simplicity, each user will be represented by a user login number below.

[0118] Next, in Figure 4 In step S102, CPU100 executes through Figure 3 The procedure for blood pressure measurement is described above. Figure 3In step S15, during the pressurization process of the cuff 20, the DC pressure change pattern is detected based on the DC component Pdc of the cuff pressure Pc. For example, Figure 7 (A) to Figure 8 (B) illustrates various pressure change patterns on the time-cuff pressure (t-Pc) plane, where the horizontal axis is time t and the vertical axis is cuff pressure Pc, showing an increasing rate of ascent and a rise (i.e., a gradual increase in slope and a downward convex curve) from the start of pressurization over time. Specifically, Figure 7 (A) shows a typical pressure change pattern PdcU1 detected for a muscular user U1 with large arms. Additionally, Figure 7 (B) shows a typical stress change pattern PdcU2 detected for a user U2 with a muscular physique and thin arms. Figure 8 (A) shows a typical stress change pattern PdcU3 detected for a user U3 with a fatty body type and thick arms. Figure 8 Figure (B) shows a typical pressure change pattern PdcU4 detected in a user U4 with a fatty body type and thin arms. Hereinafter, such DC pressure change patterns will be collectively referred to as PdcU. Additionally, in Figure 3 In step S18, during the decompression process of the cuff 20, the pulse wave signal Pm (containing the waveform pattern Pm1 for each beat) is detected as a component of the pressure change caused by the pulse wave. In this example, the CPU 100 functions as an input unit 210, inputting the acquired DC pressure change pattern PdcU to the authentication function module 200. Furthermore, in addition to the DC pressure change pattern, the waveform pattern Pm1 for each beat can also be input to the authentication function module 200 (this will be explained later).

[0119] Hereinafter, a reference numeral (e.g., U1) representing a user (or the subject being tested) is assigned to a physical quantity (e.g., Pdc), indicating that the physical quantity is obtained for that user (or the subject being tested).

[0120] Here, it is empirically known that the pressure change pattern (especially the DC pressure change pattern PdcU) depends on the subject's physical characteristics, such as the circumference of the measured area (thick or thin arm) and body composition (muscular or fatty). For example, in Figure 7 (A) and Figure 8 As shown in (A), this is the case of a thicker arm (i.e., the case where the circumference of the measured part is larger) and... Figure 7 (B) and Figure 8Compared to the case shown in (B), which is a case with a thin arm (i.e., a smaller circumference), due to the increased internal capacity of the cuff 20 (fluid bag 22), there is a tendency for the rate of increase (slope) of the cuff pressure Pc to decrease when fluid is supplied to the cuff 20 at a constant flow rate per unit time during the pressurization process. Furthermore, Figure 8 (A) and Figure 8 As shown in (B), the measured area is adipose tissue. Figure 7 (A) and Figure 7 Compared to the case of muscular physique shown in (B), the measured area tends to flatten easily in the high-pressure zone (e.g., above 25 mmHg), thus the rate of increase (slope) of cuff pressure in the high-pressure zone tends to be smaller. Moreover, these tendencies will combine depending on the combination of the circumference of the measured area, body composition, etc., of each subject (user).

[0121] Next, in Figure 4 In step S103, the CPU 100 functions as the feature registration unit 230, acquiring feature information about the DC pressure change pattern PdcU in this example. Specifically, for example, for user U1, at preset times t1 and t2 (in this example, 0 < t1 < t2 < 10 seconds) from the start of pressurization (time t = 0 seconds, cuff pressure Pc = 0 mmHg), the CPU acquires... Figure 7 The values ​​p1U1 and p2U1 obtained from the pressure change pattern PdcU1 shown in (A) are used as characteristic quantities of DC. Similarly, for user U2, values ​​obtained at times t1 and t2 are... Figure 7 The values ​​p1U2 and p2U2 obtained from the pressure change pattern PdcU2 shown in (B) are used as characteristic quantities of DC. Additionally, for user U3, values ​​obtained at times t1 and t2 are... Figure 8 The values ​​p1U3 and p2U3 obtained from the pressure change pattern PdcU3 shown in (A) are used as characteristic quantities of DC. Additionally, for user U4, values ​​obtained at times t1 and t2 are... Figure 8 The values ​​p1U4 and p2U4 obtained by the pressure change pattern PdcU4 shown in (B) are used as DC characteristic quantities. The DC characteristic quantities of these pressure change patterns PdcU (collectively referred to by the reference numerals p1 and p2) depend on the physical characteristics of the person being tested (user) as described above, and are therefore suitable as a criterion for personal authentication.

[0122] Next, in Figure 4In step S104, the CPU 100 also functions as the feature registration unit 230, establishing a correspondence between feature information (in this example, DC feature quantities p1 and p2) and user registration numbers, and storing (registering) them in the memory 51 as registration feature information. For example, if the DC feature quantities obtained for user U1 are p1U1 and p2U1, they are stored in the memory 51 as shown in Table 1 below. Furthermore, if the DC feature quantities obtained for user U2, who is different from user U1, are p1U2 and p2U2, they are distinguished from the registration feature information of user U1, and the registration feature information for user U2 is stored. The same applies to other users.

[0123] (Table 1) Login Information Form

[0124]

[0125] In this way, by performing login mode processing at the stage before personal authentication, login feature information can be stored as a benchmark for comparison of the personal authentication of the aforementioned users.

[0126] Furthermore, in this example, in Figure 4 In step S102, for a specific user, it will be done through... Figure 3 The procedure for blood pressure measurement is described as being performed in a resting state, for example, repeated at least three times. Figure 4 In step S103, the DC pressure change pattern PdcU is detected in each of three or more blood pressure measurements. Figure 4 In step S104, the characteristic information of these DC pressure change patterns PdcU (in this example, the DC characteristic quantities p1 and p2) is recorded in the registration information table. Therefore, in this example, for user U1, the DC characteristic quantities p1U1 and p2U1 each have more than three values ​​recorded. For user U2, the DC characteristic quantities p1U2 and p2U2 each have more than three values ​​recorded. The same applies to other users.

[0127] (Processing of measurement modes)

[0128] Figure 5 The flowchart illustrates the processing of a measurement mode, including blood pressure measurement based on the sphygmomanometer 1 and personal authentication. Here, the subject being authenticated (denoted by reference numeral Ux) is in a wearing state with the cuff 20 worn on the upper arm, which is the measurement site. In this state, when the subject Ux indicates the start of the measurement mode via the measurement switch 52B provided on the main body 10, the CPU 100 performs the measurement mode processing as described below.

[0129] First of all, Figure 5 In step S201, CPU100 executes through Figure 3 The procedure for blood pressure measurement is described above. Figure 3 In step S15, for the subject Ux, during the pressurization process of the cuff 20, the DC pressure change pattern (denoted by the reference numeral PdcUx) is detected as the DC component Pdc of the cuff pressure Pc. Additionally, in Figure 3 In step S18, for the subject Ux, during the decompression process of the cuff 20, the pulse wave signal Pm (containing the waveform pattern Pm1 of each beat) is detected as the pressure change component caused by the pulse wave. In this example, the CPU 100 functions as an input unit 210, inputting the DC pressure change pattern PdcUx to the authentication function module 200. Figure 3 In step S22, the CPU 100 saves the blood pressure value and pulse count in the memory 51.

[0130] Next, in Figure 5 In step S202, the CPU 100 functions as the feature acquisition unit 220, acquiring feature information about the DC pressure change pattern PdcUx in this example. Specifically, for example, at preset times t1 and t2 (in this example, 0 < t1 < t2 < 10 seconds) from the start of pressurization (time t = 0 seconds, cuff pressure Pc = 0 mmHg) (see reference). Figure 7 The value obtained by the pressure change pattern PdcUx (represented by the figure references p1Ux and p2Ux) is used as the characteristic quantity of DC.

[0131] Next, in Figure 5 In step S203, the CPU100 acts as the authentication unit 240, comparing the acquired feature information (in this example, DC feature quantities p1Ux and p2Ux) with the login feature information in the login information table of Table 1, and searching for users equivalent to the detected user Ux.

[0132] Specifically, such as Figure 9 As illustrated in the example, CPU 100 creates a feature space Q with the horizontal axis as the DC feature quantity p1 and the vertical axis as the DC feature quantity p2. In this example, in the login information table of Table 1, for user U1, the centroid point CgU1 in feature space Q is calculated using data where each of the DC feature quantities p1U1 and p2U1 has at least three logged values. Similarly, for user U2, the centroid point CgU2 in feature space Q is calculated using data where each of the DC feature quantities p1U2 and p2U2 has at least three logged values. Furthermore, the same method is used to calculate the centroid points in feature space Q for other users.

[0133] Furthermore, the CPU100 sets certain ranges as its own acceptance ranges ArU1, ArU2, ... centered on these centroids CgU1, CgU2, ... Here, "own acceptance range" refers to the permissible range within which the tested subject Ux, who should be considered as the object of authentication, is a pre-registered user U1, U2, ... In this example, each own acceptance range ArU1, ArU2, ... is set as an elliptical range with the permissible amplitude of the DC characteristic quantity p1 set as the major axis L1 and the permissible amplitude of the DC characteristic quantity p2 set as the minor axis L2. The dimensions of these own acceptance ranges ArU1, ArU2, ... are fixed values, adjusted based on experimental data obtained from multiple individuals beforehand, so that the errors of acceptance by others (the error of the self being identified as someone else) and the errors of rejection by the individual (the error of the self being identified as not being the self) are both minimized.

[0134] Then, the CPU100 determines whether the points (denoted by the figure symbol dUx) corresponding to the DC feature quantities p1Ux and p2Ux of the detected user Ux in the feature quantity space Q fall within any of the user's acceptance range ArU1, ArU2, ... . Thus, the search is performed for users equivalent to the detected user Ux. This search is performed by comparing the DC feature quantities p1Ux and p2Ux obtained by the detected user Ux with the login feature information (DC feature quantities in this example) of each user U1, U2, ...

[0135] Here, in the case where the point dUx corresponding to the DC characteristic quantities p1Ux and p2Ux of the subject Ux does not enter any of the subject's receiving range ArU1, ArU2, ... Figure 5 If step S204 is "No", CPU 100 determines that the detected user Ux is not a pre-logged-in user U1, U2, ... In this case, enter Figure 5 In step S206, the CPU 100 functions as the display processing unit 135, causing the display 50 to display only the measurement results of blood pressure and pulse rate.

[0136] On the other hand, if the point dUx corresponding to the DC characteristic quantities p1Ux and p2Ux of the subject Ux enters any one of the subject's receiving range ArU1, ArU2, ... ( Figure 5 If step S204 is "Yes", it is determined that the detected user Ux is a pre-logged-in user. In this case, Figure 5 In step S205, the CPU 100 establishes a correspondence between the blood pressure measurement result and the user's login number and saves it in the memory 51. Figure 9 In the example, the point dUx corresponding to the DC characteristic quantities p1Ux and p2Ux of the detected subject Ux enters the user U1's own reception range ArU1. Therefore, the CPU100 will... Figure 5 The blood pressure and pulse rate obtained in step S201 are correspondingly stored in memory 51 as shown in Table 2 below, corresponding to the user login number U1 of user U1. In this example, the systolic blood pressure SBP is 130 [mmHg], the diastolic blood pressure DBP is 80 [mmHg], and the pulse rate is 70 [beats / minute]. Then, proceed to... Figure 5 In step S206, the CPU 100 functions as a display processing unit 135, causing the display 50 to show the measurement results of blood pressure and pulse rate.

[0137] (Table 2) Measurement Results Table

[0138]

[0139] Alternatively, the login information table in Table 1 and the measurement results table in Table 2 can be linked together using the user login number and combined into a single table.

[0140] Thus, in this blood pressure monitor 1, the subject's personal identification is performed using characteristic information (in this example, DC characteristic quantities p1 and p2) about the pressure change pattern that increases with time from the start of inflation during the pressurization process of the cuff 20. As mentioned above, it is empirically known that this pressure change pattern (especially the DC pressure change pattern PdcU) depends on the subject's physical characteristics, such as the circumference of the measured area (thick or thin arm) and body composition (muscular or fatty body). These physical characteristics vary little over a short period of time for the individual subject. In other words, these physical characteristics differ from Korotkoff sounds or electrocardiogram signals, resulting in small deviations in each measurement. Therefore, in this blood pressure monitor 1, personal identification of the subject can be performed with high accuracy by using characteristic information about the pressure change pattern that depends on the subject's physical characteristics. Furthermore, in this case of personal identification of the subject, there is no need to install unnecessary components not used during pressure observation (e.g., sound detection devices such as microphones, electrocardiogram electrodes, etc.). Therefore, the blood pressure monitor 1 can be constructed with a simple structure and low cost.

[0141] Furthermore, in this blood pressure monitor 1, by appropriately setting the aforementioned acceptable ranges ArU1, ArU2, ..., it is possible to determine with higher accuracy whether the person being tested, Ux, who is the subject of authentication, is a pre-registered user U1, U2, ...

[0142] In the example above, although the acceptable ranges ArU1, ArU2, ... are set as elliptical ranges, this is not a limitation. For example, the acceptable ranges can also be set independently for p1 on the horizontal axis and p2 on the vertical axis, setting the acceptable ranges as rectangular ranges. Furthermore, known methods such as Euclidean distance, Manhattan distance, cosine similarity, Pearson's product-moment correlation coefficient, and dynamic time warping can be used to set the acceptable ranges.

[0143] Furthermore, in the example above, the values ​​p1 and p2 obtained from the DC pressure change pattern PdcU at two predetermined times t1 and t2 from the start of pressurization were used as characteristic quantities for DC, but this is not a limitation. For example, the values ​​p1, p2, p3, ... obtained from the DC pressure change pattern PdcU at three or more predetermined times t1, t2, t3, ... from the start of pressurization can also be used as characteristic quantities for DC. In addition, various characteristic quantities can be used as characteristic quantities for DC.

[0144] • The time Δt from the pre-set pressure p1′ to the other pressure p2′ is Δt = t2′ - t1′ (where t1′ and t2′ are the times when the pressure becomes the pre-set pressure p1′ and p2′, respectively).

[0145] • The ratio of pressures (p2 / p1);

[0146] • The slope between the pre-set times t1 and t2 is α = (p2 - p1) / (t2 - t1);

[0147] • The slope at a pre-set time

[0148] • Areas S1 and S2 (where, for example, Figure 7 As in example (A), S1 is the area from time 0 to t1 (represented by a point), and S2 is the area from time t1 to t2 (represented by a slash).

[0149] • Area ratio S2 / S1.

[0150] (Variation Example 1)

[0151] In the example above, only the DC characteristic quantity of the DC pressure change pattern PdcU is used as characteristic information, but it is not limited to this. In addition to the DC characteristic quantity, the waveform pattern Pm1 for each pulse (see reference) can also be used as characteristic information. Figure 6 The characteristic of (A) (a characteristic of one beat).

[0152] In this case, the stage prior to the personal certification of the blood pressure monitor 1 Figure 4 Step S102 (especially) Figure 3In step S18), the CPU 100 functions as an input unit 210, inputting the waveform pattern Pm1 for each pulse to the authentication function module 200, in addition to the DC pressure change pattern PdcU. Furthermore, in Figure 4 In step S103, the CPU100 functions as the feature registration unit 230, and as feature information, in addition to the DC feature quantity of the DC pressure change pattern PdcU, it acquires the feature quantity of the waveform pattern Pm1 for each pulse.

[0153] Here, Figure 10 A magnified view shows the waveform pattern Pm1 for each beat. In detail, Figure 10 This represents a typical waveform pattern Pm1 for each beat, with the start point set as t = 0 seconds, pressure = 0 mmHg, the horizontal axis as the elapsed time t from the start point, and the vertical axis as the pressure change Δp from the start point. In this example, the waveform pattern Pm1 for each beat curves upwards from the start point, becoming convex and rising, showing the first maximum (pulse peak) Max1 at time t11, the minimum Min at time t12, the second maximum (reflection peak) Max2 at time t13, and ending at time t14 with a pressure of 0 mmHg (the end point). As a characteristic quantity of such a waveform pattern Pm1 for each beat, it is possible to use... Figure 10 Examples of various features are shown below.

[0154] • Pressure change (amplitude) pMax1 at the first maximum point Max1 and pressure change (amplitude) pMax2 at the second maximum point Max2;

[0155] • The time interval ΔtMax (=t13-t11) between the first maximum point Max1 and the second maximum point Max2;

[0156] • The slope α10 after a certain time t10 from the starting point;

[0157] • The area S10 between the waveform pattern Pm1 and the horizontal axis t in each beat;

[0158] • The ratio of the amplitude pMax1 at the first maximum point Max1 to the amplitude pMax2 at the second maximum point Max2 (pMax1 / pMax2);

[0159] • The ratio of the area S11 in area S10 from time t = 0 to t12 to the area S12 in area S12 from time t = t12 to t14 (S11 / S12).

[0160] Furthermore, the waveform pattern Pm1 for each beat can be obtained from... Figure 6 Extracted from either the pressurization or depressurization process shown in (B). However, from Figure 6 As shown in (A), the waveform of each beat is most stable in the final stage of the decompression process. Therefore, for the waveform pattern Pm1 of each beat, it is preferable to extract it in the final stage of the decompression process.

[0161] Next, in Figure 4 In step S104, the CPU 100 also functions as a feature registration unit 230. In addition to the DC feature quantities (e.g., p1, p2), it also establishes a correspondence between the phase feature quantities (e.g., pMax1, pMax2) and the user registration number, and stores (registers) them as registration feature information in the memory 51. For example, if the user's registration number is U1, in addition to the DC feature quantities p1U1 and p2U1, and the amplitude pMax1U1 at the first maximum point Max1 and the amplitude pMax2U1 at the second maximum point Max2 are obtained as phase feature quantities, the information is stored in the memory 51 as shown in Table 3 below.

[0162] (Table 3) Login Information Form

[0163]

[0164] In this way, before the personal authentication of the blood pressure monitor 1, in addition to the DC characteristic quantity, the characteristic quantity of one beat can be registered as the registration characteristic information in the memory 51.

[0165] Then, during the personal certification phase of the blood pressure monitor 1 Figure 5 Step S201 (especially) Figure 3 In step S18), the CPU 100 functions as an input unit 210, inputting the waveform pattern (denoted by the reference numeral Pm1Ux) of each pulse to the authentication function module 200, in addition to the DC pressure change pattern PdcUx, for the tested object Ux. Next, in Figure 5 In step S202, the CPU 100 functions as a feature acquisition unit 220. As feature information, in addition to acquiring the DC feature quantities p1Ux and p2Ux of the DC pressure change pattern PdcUx, it also acquires the one-beat feature quantity of the waveform pattern Pm1Ux for each beat. In this example, it acquires the amplitude pMax1 (denoted by the reference numeral pMax1Ux) at the first maximum point Max1 and the amplitude pMax2 (denoted by the reference numeral pMax2Ux) at the second maximum point Max2. Next, in... Figure 5In step S203, the CPU 100 functions as the authentication unit 240, comparing the acquired feature information (in this example, the DC feature quantities p1Ux, p2Ux and the one-step feature quantities pMax1Ux, pMax2Ux) with the login feature information in the login information table of Table 2, and searching for users equivalent to the detected user Ux. For example, the CPU 100 creates a four-dimensional feature space (denoted by reference numeral Q1) with p1, p2, pMax1, and pMax2 as coordinate axes. In the feature space Q1, points corresponding to the login feature information are calculated for pre-registered users U1, U2, ..., and an acceptable range (acceptable range) is set around the points corresponding to the login feature information to determine whether the detected user Ux is a pre-registered user. Then, when a point in the feature space Q1 corresponding to the feature information obtained for the subject Ux falls within the allowable range, the CPU 100 determines that the subject Ux is a pre-registered user (e.g., U1). Conversely, when a point corresponding to the feature information obtained for the subject Ux does not fall within the allowable range, the CPU 100 determines that the subject Ux is not a pre-registered user U1, U2, ... In this way, in addition to the DC feature quantities, a single-step feature quantity is also used as the judgment material, thereby enabling more accurate personal authentication of the subject.

[0166] Furthermore, under the pre-set first condition, in order to perform personal authentication of the subject, it is preferable to use only the DC characteristic quantity instead of the one-time characteristic quantity. Here, the "pre-set first condition" can be, for example, a condition where the reliability of the one-time characteristic quantity is low, as follows.

[0167] • The pulse count during personal authentication is significantly higher or lower than the pulse count when logging in with the above login feature information;

[0168] • The blood pressure value during personal authentication is significantly higher or lower than the blood pressure value when logging in with the above login feature information;

[0169] • The temperature, humidity, and air pressure during personal authentication are significantly higher or lower than the temperature, humidity, and air pressure when logging in the above-mentioned login feature information (in addition, in this case, in sphygmomanometer 1, temperature, humidity, and air pressure are observed and recorded for each blood pressure measurement).

[0170] Given the low reliability of one-shot feature quantities, in order to perform personal authentication of the tested person, only DC feature quantities are used instead of one-shot feature quantities, thereby preventing a decrease in the accuracy of personal authentication.

[0171] (Variation Example 2)

[0172] In the example above, to perform personal authentication on the subject, CPU100 creates a feature space Q (or Q1), searching within feature space Q for users equivalent to the subject Ux, but not limited to this. For example, blood pressure monitor 1 has features such as... Figure 11 The neural network NN1 shown in (B) after learning can also be used to determine the user equivalent to the detected person Ux.

[0173] Specifically, such as Figure 11 As shown in (B), the neural network NN1 has an input layer IL1, three intermediate layers ML1, ML2, and ML3, and an output layer OL1. In this example, the input layer IL1 and the output layer OL1 each include five nodes (neurons). The intermediate layers ML1, ML2, and ML3 each include 20 to 30 nodes.

[0174] In this example, during the learning phase, the neural network NN1 weights the coupling strength between nodes nd and nd, such that when any one of the feature information inUa, inUb, inUc, inUd, and inUe of the DC pressure change pattern PdcU for five teacher users (denoted by the appended reference numerals Ua, Ub, Uc, Ud, and Ue) is input to the input layer IL1, the output layer OL1 is output via intermediate layers ML1, ML2, and ML3, representing the corresponding teacher user Ua, Ub, Uc, Ud, or Ue. That is, the neural network NN1 has completed its learning.

[0175] Specifically, the characteristic information inUa regarding the pressure change pattern of teacher users (e.g., Ua) is preset as follows: Figure 12 As shown, in Figure 4 Step S102 (especially) Figure 3 In step S18), the DC pressure change pattern obtained (denoted by the reference numeral PdcUa) has the following values ​​p1Ua, p2Ua, p3Ua, p4Ua, and p5Ua at pre-set times t1, t2, t3, t4, and t5 (in this example, 0 < t1 < t2 < t3 < t4 < t5) from the start of pressurization (time t = 0 seconds, cuff pressure Pc = 0 mmHg). Using the same reference numerals as the feature information inUa, the vector composed of these five values ​​(p1Ua, p2Ua, p3Ua, p4Ua, p5Ua) is called the feature vector inUa. Similarly, as... Figure 11As shown in (A), the feature information inUb of the DC pressure change pattern PdcUb for teacher user Ub is represented by the feature vector inUb = (p1Ub, p2Ub, p3Ub, p4Ub, p5Ub). The feature information inUc of the DC pressure change pattern PdcUc for teacher user Uc is represented by the feature vector inUc = (p1Uc, p2Uc, p3Uc, p4Uc, p5Uc). The feature information inUd of the DC pressure change pattern PdcUd for teacher user Ud is represented by the feature vector inUd = (p1Ud, p2Ud, p3Ud, p4Ud, p5Ud). Furthermore, the feature information inUe of the DC pressure change pattern PdcUe for teacher user Ue is represented by the feature vector inUe = (p1Ue, p2Ue, p3Ue, p4Ue, p5Ue). During the learning phase, these feature vectors inUa, inUb, inUc, inUd, and inUe are input into the input layer IL1.

[0176] On the other hand, the output information teUa represents the teacher user (e.g., Ua). Figure 11 As shown in (C), the output vector teUa = (1, 0, 0, 0, 0) represents the information. Similarly, the output information teUb representing teacher user Ub is represented by the output vector teUb = (0, 1, 0, 0, 0). The output information teUc representing teacher user Uc is represented by the output vector teUc = (0, 0, 1, 0, 0). The output information teUd representing teacher user Ud is represented by the output vector teUd = (0, 0, 0, 1, 0). The output information teUe representing teacher user Ue is represented by the output vector teUe = (0, 0, 0, 0, 1). During the learning phase, for example, when the feature vector inUa of teacher user Ua is input to the input layer IL1, the corresponding output vector teUa representing teacher user Ua is provided to the output layer OL1 as the teacher signal. Similarly, when the feature vector inUa of teacher user Ua is input to the input layer IL1, the corresponding output vector teUa representing teacher user Ua is provided to the output layer OL1 as a teacher signal. Likewise, when the feature vector inUb of teacher user Ub is input to the input layer IL1, the corresponding output vector teUb representing teacher user Ub is provided to the output layer OL1 as a teacher signal. The same processing is performed for other teacher users Uc, Ud, and Ue.

[0177] By repeating this learning process, the neural network NN1 weights the coupling strength between each node nd, so that when any one of the feature information inUa, inUb, inUc, inUd, inUe of the DC pressure change pattern PdcU of the five teacher users Ua, Ub, ..., Ue is input to the input layer IL1, the output information teUa, teUb, teUc, teUd, or teUe corresponding to the input feature information is output to the output layer OL1 via the intermediate layers ML1, ML2, and ML3.

[0178] By employing such a neural network NN1, it becomes possible to perform individual authentication on user groups, including the aforementioned five teacher users Ua, Ub, ..., Ue, and / or users other than teacher users. Next, the processing of login patterns and determination patterns when using this neural network NN1 will be explained.

[0179] (Login pattern processing when using neural networks)

[0180] In this case, the CPU100 functions as the feature registration unit 230, performing the following control in advance before personal authentication. First, during the pressurization process of the cuff 20, for the user group including the five teachers Ua, Ub, ..., Ue and / or users other than teachers, for each user, as feature information about the DC pressure change pattern PdcU of that user (denoted by the attached figure Uy), the feature vector inUy = (p1Uy, p2Uy, p3Uy, p4Uy, p5Uy) is obtained. Next, for each user, the feature vector inUy obtained for that user's Uy is input to the input layer IL1 of the neural network NN1, obtaining information fe1Uy, fe2Uy, ..., feNUy presented in one of the multiple intermediate layers ML1, ML2, ML3 (in this example, the intermediate layer ML3 before the output layer OL1). The vector composed of these N values ​​(N = 20-30) is called the feature AI (Artificial Intelligence) vector feUy for user Uy. Furthermore, in Figure 11In (B), the components (fe1, fe2, ..., feN) of a general feature AI vector fe, where the user is not specified, are shown. Each component fe1, fe2, ..., feN is referred to as an AI feature quantity. It is well known that the feature AI vector fe = (fe1, fe2, ..., feN) can be used as feature information for a specific user, therefore a detailed explanation is omitted. As shown in the login information table in Table 4 below, the CPU 100 stores (logs in) the components of the feature AI vector feUy for user Uy as login feature information corresponding to user Uy in memory 51. Similarly, for users different from the aforementioned user Uy, whose login ID is Uz, if the obtained feature AI vector is inUz = (fe1Uz, fe2Uz, ..., feNUz), it is distinguished from the login feature information of user Uy, and the login feature information for user Uz is stored. The same applies to other users. Therefore, as shown in the login information table in Table 4 below, the login feature information for each user Uy, Uz, ... is logged in the memory 51.

[0181] (Table 4) Login Information Form

[0182]

[0183] (Processing of measurement patterns when using neural networks)

[0184] During the personal authentication phase, the CPU100 functions as the feature acquisition unit 220, acquiring feature information about the pressure change pattern PdcU of the person being authenticated (let's call it Ux) during the pressurization process of the cuff 20 (in this example, the feature vector inUx = (p1Ux, p2Ux, p3Ux, p4Ux, p5Ux)).

[0185] Furthermore, the CPU 100, acting as the authentication unit 240, inputs the feature AI vector inUx obtained for the detected user Ux into the input layer IL1 of the neural network NN1, obtaining the information presented in the intermediate layer ML3 as the feature AI vector inUx = (fe1Ux, fe2Ux, ..., feNUx)). Then, the CPU 100 compares the feature AI vector inUx = (fe1Ux, fe2Ux, ..., feNUx) obtained for the detected user Ux with the login feature information of each user Uy, Uz, ... in the login information table in Table 4, determining whether the detected user Ux is a user included in the user group. Thus, if the detected user Ux is a user included in the user group, this fact can be authenticated.

[0186] (Variation Example 3)

[0187] Under a pre-defined second condition, it is preferable to update the login feature information in Tables 1, 3, and 4 of the login information stored in the memory 51. Here, the "pre-defined second condition" may be, for example, a condition under which the above-mentioned login feature information should be updated.

[0188] • A certain period of time has elapsed since the login of the above-mentioned login feature information (or the last update) (e.g., a period of six months during which the feature information of the above-mentioned pressure change pattern may change significantly);

[0189] Although the detected user Ux is determined to be a pre-logged-in user (e.g., U1), the point dUx corresponding to the feature information obtained for the detected user Ux (refer to...) Figure 9 Within my acceptable range ArU1, away from the center of gravity CgU1.

[0190] Therefore, in this blood pressure monitor 1, the CPU 100 functions as an update unit 250. When it is determined that the person being tested, Ux, is a pre-registered user, the CPU uses the characteristic information obtained for the person being tested, Ux, to update the user's registration characteristic information under a pre-set second condition. Thus, when the registration characteristic information should be updated, it can be automatically updated and maintained in an appropriate state. As a result, the accuracy of the personal authentication can be prevented from decreasing.

[0191] (Variation Example 4)

[0192] Under a pre-defined third condition, it is preferable to suspend personal authentication. Here, the "pre-defined third condition" may be, for example, a condition that is considered to have low reliability of characteristic information (characteristic quantities of DC and / or characteristic quantities of a single beat) regarding the pressure change pattern.

[0193] • An irregular pulse wave was detected (in this case, blood pressure monitor 1 detects the irregular pulse wave based on the pulse rate);

[0194] • Determine if there is body movement (in this case, the blood pressure monitor 1 is equipped with an accelerometer to detect body movement);

[0195] • Determined as loose wrapping (the cuff 20 is loosely wrapped relative to the measurement site) (In this case, the sphygmomanometer 1 detects loose wrapping based on the very slow rise in the pressure change pattern)

[0196] Therefore, in this blood pressure monitor 1, the CPU 100 functions as the authentication unit 240, suspending personal authentication under a pre-set third condition. This allows for the suspension of personal authentication when the reliability of characteristic information regarding the pressure change pattern (characteristic values ​​of DC and / or characteristic values ​​of a single beat) is low. As a result, it prevents a decrease in the accuracy of personal authentication.

[0197] Furthermore, in the above-described embodiment, the measurement site is the upper arm, but it is not limited to this. The measurement site can be the upper limb other than the upper arm, such as the wrist, or the lower limb, such as the ankle.

[0198] The above embodiments are merely examples, and various modifications can be made without departing from the scope of the invention. The multiple embodiments described can be implemented independently or combined with each other. Furthermore, features in different embodiments can be implemented independently or combined with each other.

[0199] Explanation of reference numerals in the attached figures

[0200] 1. Blood pressure monitor

[0201] 10 main bodies

[0202] 20 Blood Pressure Measurement Cuff

[0203] 31 pressure sensor

[0204] 51 memory

[0205] 100 CPUs

[0206] 311 First Filter Section

[0207] 315 Second Filter Section

Claims

1. A blood pressure monitor comprising a cuff for compressing a measurement site on a subject, and measuring blood pressure by observing the pressure of the cuff, characterized in that, have: The pressure control unit controls the supply of fluid to the cuff to increase pressure or the discharge of fluid from the cuff to decrease pressure. The pressure detection unit detects the pressure of the cuff; The blood pressure calculation unit calculates blood pressure based on the output of the pressure detection unit; The feature acquisition unit acquires feature information about the pressure change pattern of the cuff over time from the start of pressure application for the tested subject who is being authenticated. The feature information includes feature information about the DC pressure change pattern, which corresponds to the DC component of the cuff pressure and the rate of increase increases over time. as well as The authentication department compares the acquired feature information with the login feature information of pre-logged-in users to perform personal authentication on the person being tested.

2. The blood pressure monitor as described in claim 1, characterized in that, The authentication department creates a feature space where the different feature quantities contained in the feature information about the pressure change pattern are set as coordinate axes. In the feature space, points corresponding to the login feature information are calculated for the pre-logged-in user, and an allowable range is set around the points corresponding to the login feature information to identify the detected person as the pre-logged-in user. In the feature space, when a point corresponding to the feature information obtained for the detected person enters the allowable range, the detected person is determined to be the pre-logged-in user. On the other hand, when a point corresponding to the feature information obtained for the detected person does not enter the allowable range, the detected person is determined not to be the pre-logged-in user.

3. The blood pressure monitor as described in claim 1 or 2, characterized in that, have: The first pressure detection unit extracts the DC component from the pressure of the cuff during the pressurization process of the cuff and detects the DC pressure change pattern. as well as The second pressure detection unit extracts the pressure variation component caused by the pulse wave at the measured location from the pressure of the cuff during the pressurization process of the cuff or during the depressurization process after the pressurization process, and detects the waveform pattern of each beat. In addition to acquiring the DC feature quantity related to the DC pressure change pattern, the feature acquisition unit also acquires the feature quantity of each beat of the waveform pattern, as the feature information. In addition to comparing the acquired DC characteristic values ​​with the pre-registered DC characteristic values, the authentication department also compares the acquired one-shot characteristic values ​​with the pre-registered one-shot characteristic values ​​to perform personal authentication for the person being tested.

4. The blood pressure monitor as described in claim 3, characterized in that, Under a pre-set first condition, the authentication department uses only the DC characteristic quantity instead of the one-time characteristic quantity in order to perform personal authentication of the tested person.

5. The blood pressure monitor as described in claim 1 or 2, characterized in that, have: Storage department; and In the stage prior to the personal authentication, the feature login unit performs the following control in advance: during the compression of the cuff, it acquires feature information about the pressure change pattern of the subject, establishes a correspondence between the acquired feature information and the subject, and stores it in the storage unit as the login feature information.

6. The blood pressure monitor as described in claim 1 or 2, characterized in that, have: Storage Department: and A neural network has an input layer containing multiple nodes, multiple intermediate layers, and an output layer. The neural network learns by weighting the nodes, so that when any one of the feature vectors representing feature information about the pressure change patterns of multiple teacher users is input into the input layer, the output layer outputs the output information of the teacher user corresponding to the input feature information through the multiple intermediate layers. The blood pressure monitor has a feature login unit. Before performing the personal authentication, this feature login unit performs the following pre-control: During the inflation of the cuff, for a user group including multiple teacher users and / or users other than teacher users, for each user, it acquires feature information about the pressure change pattern of that user. The acquired feature information is then input as a feature vector into the input layer of the neural network, so that information from one of the multiple intermediate layers corresponds to the user and is stored in the storage unit as the login feature information. During the personal authentication phase, the feature acquisition unit acquires feature information about the pressure change pattern of the subject being authenticated during the compression of the cuff. The authentication unit inputs the acquired feature information as a feature vector into the input layer of the neural network, compares the information of a certain intermediate layer presented in the plurality of intermediate layers with the login feature information of each user in the user group logged in to the storage unit, and determines whether the detected person is a user included in the user group.

7. The blood pressure monitor as described in claim 5, characterized in that, have: When the updating unit determines that the detected user is a pre-logged-in user, it uses the feature information obtained for the detected user and updates the user's login feature information under a pre-set second condition.

8. The blood pressure monitor as described in claim 1 or 2, characterized in that, The authentication department suspends the personal authentication under a pre-set third condition.

9. A personal authentication method in a blood pressure monitor, characterized in that, The blood pressure monitor is the blood pressure monitor according to claim 1. In the aforementioned personal authentication method, Using the feature acquisition unit, for the subject being authenticated, during the compression of the cuff, feature information about the pressure change pattern of the cuff over time from the start of compression is acquired. This feature information includes feature information about a DC pressure change pattern, which corresponds to the DC component of the cuff pressure, and the rate of increase increases over time. Using the authentication unit, the acquired feature information is compared with the login feature information of the pre-logged-in user to perform personal authentication on the person being detected.

10. A storage component storing a program that causes a computer to execute the personal authentication method in the blood pressure monitor of claim 9.

Images