Apparatus, method and procedure for estimating serum triglycerides

Abstract

Conventional methods for measuring blood triglycerides involve invasive blood sampling, such as inserting needles into the skin, which can cause psychological or physiological burdens on the subject. According to the present invention, a blood triglyceride estimation model is generated through machine learning based on prior information about the attributes of several subjects, non-invasive biological information, and blood test data. This model allows for the non-invasive estimation of blood triglyceride levels based on the user's attribute information and / or non-invasive biological information.

Description

[Technical Field]

[0001] This invention relates to a device, method and program for estimating blood triglycerides. [Previous Technology]

[0002] It is known that blood triglycerides are generally measured by taking blood samples from the subject. However, this method requires invasive procedures such as inserting needles into the subject's skin, which can cause psychological or physiological burden on the subject. As a non-invasive method for estimating blood triglycerides, Patent Document 1 describes a method for estimating blood triglycerides by irradiating the fingertip with light of a specific wavelength. [Prior Art Documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent No. 6241853 [Non-Patent Document]

[0004] [Non-Patent Literature 1] Psychology Research and Behavior Management 2011:4 81-86, Summary of the clinical investigations ES Teck Complex March, 20, 2010 Abstract [Non-Patent Literature 2] RN Chua, YW Hau, CM Tiew and WL Hau, "Investigation of Attention Deficit / Hyperactivity Disorder Assessment Using Electro Interstitial Scan Based on Chronoamperometry Technique," in IEEE Access, vol. 7, pp. 144679-144690, 2019, doi: 10. 1109 / ACCESS. 2019. 2938095. [Non-Patent Literature 3] Maarek A. Electro interstitial scan system: assessment of 10 years of research and development. Med Devices (Auckl). 2012;5:23-30. doi:10.2147 / MDER.S29319 [Summary of the Invention]

[0005] (The problem that the invention aims to solve)

[0006] The object of the present invention is to provide a device, method, and program for estimating blood triglycerides, which, in order to non-invasively and accurately estimate blood triglycerides, can estimate blood triglycerides based not only on pulse data obtained by irradiating light of a specific wavelength, but also on body mass index (BMI), blood pressure, electrocardiogram data, and bioimpedance. (Technical means to solve the problem)

[0007] The blood triglyceride estimation device of the present invention is characterized by comprising: an information acquisition unit that acquires attribute information and non-invasive biological information of a predetermined user; an estimation model memory unit that stores a blood triglyceride estimation model; and an estimation processing unit that uses the blood triglyceride estimation model to calculate the estimated blood triglyceride value of the predetermined user based on the attribute information and / or non-invasive biological information of the predetermined user.

[0008] The blood triglyceride estimation device is characterized by further having: a training data memory unit that stores a training data set; and a learning processing unit that generates a blood triglyceride estimation model based on the training data set by means of machine learning.

[0009] The blood triglyceride estimation device is characterized in that: the attribute information includes either age and sex, or a combination thereof, and the non-invasive biological information includes BMI, blood pressure, pulse wave data, electrocardiogram data and biological impedance, or a combination thereof.

[0010] The blood triglyceride estimation device is characterized in that: the training dataset includes the subject’s attribute information, non-invasive biological information and blood triglyceride measurement values ​​obtained from the blood.

[0011] The characteristic of the blood triglyceride estimation device is that the correlation coefficient between the logarithm of the blood triglyceride estimation value and the logarithm of the blood triglyceride measurement value is 0.6 or higher.

[0012] The blood triglyceride estimation device is characterized in that: the estimation processing unit calculates the blood triglyceride risk estimation value, rather than calculating the blood triglyceride estimation value.

[0013] The blood triglyceride estimation device is characterized in that: the learning processing unit adds labels to the training dataset indicating the presence or absence of blood triglyceride risk based on blood triglyceride measurement values ​​obtained from blood, and when the difference between the number of blood triglyceride risk and the number of blood triglyceride risk is greater than a predetermined value, the sample data in the training dataset is increased to reduce the difference.

[0014] The blood triglyceride estimation device is characterized in that: the learning processing unit generates a first blood triglyceride risk estimation model and a second blood triglyceride risk estimation model respectively by machine learning based on different types of training data sets, and the estimation processing unit uses the first blood triglyceride risk estimation model and the second blood triglyceride risk estimation model to calculate the blood triglyceride risk estimation value of a given user.

[0015] The blood triglyceride estimation device is characterized by having a biological information estimation unit that estimates at least one of the biological information included in biological information, such as BMI, blood pressure, pulse wave data, electrocardiogram data and biological impedance, and an information acquisition unit that acquires the biological information estimated by the biological information estimation unit as the biological information of a predetermined user.

[0016] The blood triglyceride estimation device is characterized in that, in addition to the blood triglyceride estimation device, it further has a biological information measuring device for measuring non-invasive biological information, thus constituting a non-invasive blood triglyceride estimation system.

[0017] The method for estimating blood triglycerides in this invention is characterized by comprising the following steps: memorizing a training data set containing the subject's attribute information, non-invasive biological information, and blood triglyceride measurement values ​​obtained from the blood; generating a blood triglyceride estimation model based on the training data set by machine learning; and using the blood triglyceride estimation model, calculating the estimated blood triglyceride value of the predetermined user based on the predetermined user's attribute information and / or non-invasive biological information.

[0018] The program in this invention is characterized by enabling a computer to perform the following steps: memorizing a training dataset containing the subject's attribute information, non-invasive biological information, and blood triglyceride levels measured from blood; generating a blood triglyceride estimation model based on the training dataset using machine learning; and using the blood triglyceride estimation model, calculating the estimated blood triglyceride value for a given user based on the user's attribute information and / or non-invasive biological information. (Effectiveness compared to prior art)

[0019] According to the present invention, non-invasive biological information can be used to estimate triglycerides in bleeding or the risk of triglycerides in bleeding with excellent accuracy through machine learning.

Implementation Method

[0021] Hereinafter, embodiments will be described with reference to drawings. Furthermore, these embodiments are illustrative only, and the present invention is not limited to the following configuration.

[0022] <Function of the Device> Referring to Figures 1 to 6, the blood triglyceride estimation system 1 and blood triglyceride estimation device 30 of this embodiment will be described. Figure 1 is a block diagram showing the general structure of the blood triglyceride estimation system 1 of this embodiment. The blood triglyceride estimation system 1 includes a terminal device 10, a biological information measuring device 20, a blood triglyceride estimation device 30, and a display device 39. Here, "user" refers to a person who uses the blood triglyceride estimation system to obtain a non-invasive blood triglyceride or its risk estimation value. Furthermore, "subject" refers to a person who, after completing the prescribed procedures and obtaining consent, provides age and gender attribute information, non-invasive biological information, and blood triglyceride measurement values ​​obtained from blood as a training dataset used in the blood triglyceride estimation system.

[0023] The terminal device 10 can be any information terminal that allows input of user attribute information (name, ID, age, gender, etc.) and can output the input information to the blood triglyceride estimation device 30 via a wired or wireless communication network. Examples include mobile terminals such as tablets, smartphones, and wearable devices, or personal computers (PCs). Furthermore, height, weight, etc., can be measured using the biometric information measurement device 20 described below.

[0024] The biometric information measuring device 20 measures the user's non-invasive biometric information. Here, non-invasive biometric information refers to biometric information obtained by means of methods that do not require the insertion of instruments into the skin or body openings. Non-invasive biometric information can be measured using commercially available height and weight scales, blood pressure monitors, pulse oximeters, pulse wave meters, electrocardiographs, impedance meters, and skin conductance meters. Alternatively, the Estec BC-3 (dual-purpose system), which can simultaneously measure pulse wave data, electrocardiogram data, and biometric impedance, can also be used. These devices can measure non-invasive biometric data without causing psychological or physiological burden to the user.

[0025] In an embodiment of the present invention, non-invasive biological information includes any one or a combination thereof, such as BMI (body mass index), blood pressure, pulse wave data, electrocardiogram data, and biological impedance.

[0026] BMI is calculated based on height h[m] and weight w[kg] using the following formula: BMI=w / h2[kg / m2].

[0027] Blood pressure includes any one of systolic blood pressure, diastolic blood pressure, pulse pressure, and mean arterial pressure, or a combination thereof. Here, pulse pressure is calculated by the following formula: Pulse pressure = Systolic blood pressure - Diastolic blood pressure. And mean arterial pressure is calculated by the following formula: Mean arterial pressure = Diastolic blood pressure + Pulse pressure × 1 / 3.

[0028] Pulse data is measured using a pulse oximeter or pulse wave meter by irradiating red light (~660 nm) onto a protruding part of the body, such as the fingers, with a red light-emitting diode (LED) or a near-infrared light (~905 nm) irradiated with an infrared (IR) LED, and then measuring the transmitted light using a photoelectric crystal. Pulse data includes pulse rate, elasticity index, peripheral vascular resistance, acceleration pulse wave, b / a, e / a, -d / a, Gauzer acceleration pulse wave aging index, ejection rate, left ventricular ejection time (LVET), dicrotic-pulse elasticity index (DEI), or a combination thereof. Here, the elasticity index refers to the value obtained by dividing height by the time from the onset of the systolic peak of the fingertip volume pulse wave to the detection of its diastolic peak. Peripheral vascular resistance is calculated by dividing mean arterial pressure by cardiac output × 80. The dilatational elasticity index (DEI) is an indicator of diastolic vascular elasticity and can be measured using a pulse wave velocity (PWV) measuring device. A value of 0.3–0.7 is considered normal; below 0.3 suggests the possibility of hypertension or arteriosclerosis; and above 0.7 suggests the possibility of acute anxiety neurosis. Accelerated pulse wave is the second derivative of photoplethysmogram (SDPTG). The accelerated pulse wave includes an initial positive wave (a wave), an initial negative wave (b wave), a mid-systolic rising wave (c wave), a late-systolic falling wave (d wave), and an initial diastolic positive wave (e wave). The b / a, e / a, and -d / a ratios are calculated based on the proportions of each wave height. With age, the ratio of b / a increases while the ratios of c / a, d / a, and e / a decrease. Therefore, the Gaussian acceleration pulse wave aging index (bcde) / a can be used to assess vascular aging. Ejection rate is proportional to the acceleration pulse wave aging index, which is determined by the proportion of blood pumped from the ventricles with each heartbeat. LVET is the left ventricular ejection time, the time it takes for blood to be ejected from the left ventricle into the aorta after the aortic valve opens.

[0029] Electrocardiogram (ECG) data can be measured using electrocardiography (ECG) with electrodes or photoplethysmography (PPG). ECG data includes respiratory rate, heart rate, RR interval, standard deviation of RR interval, MxDMn ratio, power spectrum in the low-frequency band, power spectrum in the high-frequency band, heart rate variability index (Low Frequency / High Frequency), total power, or a combination thereof. Here, the RR interval refers to the interval from one QRS wave to the next on the ECG. The MxDMn ratio is the ratio of the longest to the shortest RR interval over a given time period, representing an irregular heartbeat index. Total power refers to the calculated total power of the power spectrum at frequencies of 0–0.4 Hz (Very Low Frequency, LF, HF) during a 2-minute measurement. This value reflects the overall activity of the autonomic nervous system, which is dominated by sympathetic activity. By calculating the power spectral density based on the electrocardiogram, the power spectral density ratio of high frequency (0.1875-0.50 Hz: HF), the power spectral density ratio of low frequency (0.05-0.1875 Hz: LF), the LF / HF ratio, and the power spectral density ratio of ultra-low frequency band (0-0.05 Hz: VLF) can be calculated.

[0030] The impedance (conductance) of a living organism can be measured by passing a weak current between two of six electrodes, such as those on both feet, both hands, and the left and right foreheads. If a current is passed between two of the six electrodes, the following can be measured: (1) the conductance of the anode / cathode (μS), (2) the conductance of the cathode / anode (μS), (3) the difference between the conductance measured by (1) and the conductance measured by (2) (∆SCRA-SCRC), and (4) the conductivity (μS / m). Furthermore, muscle mass, body fat mass, total water content, phase angle, and resistance value can also be measured simultaneously. Moreover, the dielectric constant (μSi) when current flows between the right and left hands, and between the right and left foreheads, can also be measured. Preferably, the impedance (conductance) of a living organism is measured using the electrical conductivity of 22 patterns from the six electrodes.

[0031] Biological impedance includes body fat mass (kg), body fat mass (%), lean body mass, lean body percentage, muscle mass, total water content (kg), total water content (%), intracellular water content (%), cardiac output, 1 frontal left-2 right-hand / SCR A, 1 frontal left-2 right-hand / ∆SCR C-SCR A, 5 left-hand - 6 left-foot / ∆SCR C-SCR A, 13 left-foot - 14 right-foot / SCR A, 15 right-hand - 16 frontal left / ∆SCR C-SCR A, 15 right-hand - 16 frontal left SCR C, 19 right-foot - 20 left-hand / ∆SCR C-SCR A. ESG2+4+15+17 (μS / m), ESG6+13+19 (%), ESG6+8+19+21 (%), ESG6+8+19+21 (μS / m), ESG9+10 (μS / m), ESG9+10 (%), conductance of left foot, R (Ω), phase angle, dielectric constant of frontal path, conductivity of frontal path (9), dielectric constant of single-hand-single-hand path, conductivity of hand to hand (11, 12), any of the following: stroke volume (cardiac output ÷ heart rate), or a combination thereof.

[0032] Here, SCR is an abbreviation for skin conductance response, and ESG is an abbreviation for electron scanning spectroscopy. The "+" in ESG2+4+15+17 indicates which electrode on the body was used to measure the conductance. For example, ESG2+4+15+17 represents the average conductance measured as shown in Figure 2 using the left hand when an electric current is applied from the left hand to the left forehead, the right hand when an electric current is applied from the right hand to the right forehead, the right hand when an electric current is applied from the right hand to the left forehead, and the left hand when an electric current is applied from the left hand to the right forehead. Such conductances are described in detail in Non-Patent Document 1. Furthermore, the method for measuring ESG (electron scanning spectroscopy) is described in detail in Non-Patent Documents 2 and 3. "1 frontal left - 2 right hand / SCR A" refers to the conductivity (or conductivity) of the path when current flows through "1 frontal left" as the cathode and "2 right hand" as the anode. "5 left hand - 6 left foot / ∆SCR C - SCR A" refers to the difference in conductivity when current flows through "5 left hand" and "6 left foot" in an anode-cathode or cathode-anode configuration.

[0033] BMI and blood pressure can be measured using a height and weight scale and a blood pressure monitor.

[0034] The non-invasive biological information measured is output to the blood triglyceride estimation device 30 via a wired or wireless communication network. The biological information measuring device 20 can be a built-in measuring device or a portable measuring device such as a wearable terminal.

[0035] The blood triglyceride estimation device 30 includes a first acquisition unit 31, a second acquisition unit 32, a user data memory unit 33, a training data memory unit 34, a learning processing unit 35, an estimation model memory unit 36, an estimation processing unit 37, and an estimation data memory unit 38. The first acquisition unit 31 acquires the user's attribute information from the terminal device 10. Furthermore, the second acquisition unit 32 acquires the user's non-invasive biological information from the biological information measurement device 20.

[0036] The user data memory unit 33 stores the user's attribute information and non-invasive biological information obtained from the first acquisition unit 31 and the second acquisition unit 32.

[0037] The training data memory unit 34 stores several training data sets containing attribute information of several subjects obtained in advance, non-invasive biological information, and specimen examination information (measured values) such as blood triglycerides obtained through blood tests, and uses these as training data sets for machine learning. Furthermore, the specimen examination information may also include examination information obtained from blood, urine, stool, etc.

[0038] The learning processing unit 35 acquires the training data set stored in the training data memory unit 34 and uses the training data set to create a blood triglyceride estimation model. Specifically, the acquired training data set is normalized, and through machine learning such as neural networks (NN), XGBoost, and linear regression, the relationship between attribute information and non-invasive biological information and blood triglyceride levels obtained through blood tests is learned, generating an estimation model for blood triglyceride levels based on non-invasive biological information. The estimation model memory unit 36 ​​stores the blood triglyceride estimation model generated by the learning processing unit 35. The non-invasive biological information includes any one or a combination of BMI, blood pressure, pulse wave data, electrocardiogram data, and biological impedance.

[0039] The estimation processing unit 37 uses the estimation model generated by the learning processing unit 35 to estimate the user's blood triglyceride level based on the user's attribute information and / or non-invasive biological information. Then, the estimated blood triglyceride value is stored in the estimation data storage unit 38.

[0040] The display device 39 can display the estimated value of triglycerides in the blood along with the user's attribute information and non-invasive biological information. Furthermore, such data can also be displayed on the terminal device 10 held by the user.

[0041] <Hardware Configuration of the Device> Figure 3 is a hardware configuration diagram of the blood triglyceride estimation device 30. As shown in Figure 3, the blood triglyceride estimation device 30 is composed of a computer 300 having one or more processors 301, memory 302, storage 303, input / output ports 304, and communication ports 305. The processor 301 performs processing related to the blood triglyceride estimation of this embodiment by executing a program. The memory 302 temporarily stores the program and the results of the program's calculation. The storage 303 stores the program for the blood triglyceride estimation device 30 to execute. The storage 303 can be any storage device as long as it can be read by a computer, such as various recording media (magnetic disks, optical disks, etc.), random access memory, flash memory, read-only memory, etc. Input / output port 304 receives information from terminal device 10 and biological information measuring device 20, or outputs estimated blood triglyceride values ​​to display device 39. Communication port 305 transmits and receives data with other information terminals such as computers (not shown). Communication can be wireless or wired. Furthermore, the blood triglyceride estimation device 30 can be implemented using a commercially available desktop or laptop PC, and the time required to calculate the estimated triglyceride value using the estimation model is a few seconds. Moreover, when the processor 301 of the blood triglyceride estimation device 30 operates, the aforementioned first acquisition unit 31, second acquisition unit 32, learning processing unit 35, and estimation processing unit 37 function.

[0042] <Generating a Blood Triglyceride Estimation Model via Machine Learning> Figure 4 is a flowchart showing the execution steps of generating a blood triglyceride estimation model via machine learning. In step ST101, the learning processing unit 35 preprocesses the input data (e.g., the training dataset mentioned above). Specifically, the learning processing unit 35 converts the gender attribute information of each subject into a one-hot vector. Furthermore, the learning processing unit 35 normalizes the attribute information other than gender, the non-invasive biological information of each subject, and the logarithm of the blood triglyceride measurement values ​​obtained by blood tests using Yeo-Johnson transformation or Bocc-Box transformation, etc. In step ST102, the learning processing unit 35 performs machine learning in parallel using linear regression, neural networks (NN), and gradient boosting regression trees. Furthermore, any one of linear regression, neural networks (NN), and gradient boosting regression trees, or a combination of two, can be selected. Furthermore, grid search can also be used for hyperparameter tuning of gradient boosting. When performing machine learning using gradient boosting regression trees, software libraries such as XGBoost, CatBoost, and LightBGM can be used. In step ST103, the learning processing unit 35 performs unsampled learning (ridge regression) on the learning results obtained through linear regression, neural networks (NN), and gradient boosting regression trees. Moreover, if any of linear regression, neural networks (NN), or gradient boosting regression trees is selected in step ST102, unsampled learning (ridge regression) in step ST103 is not required. The learning processing unit 35 stores the blood triglyceride estimation model generated through the above learning processing in the estimation model memory unit 36.

[0043] When performing machine learning using linear regression, for example, linear regression provided by the open-source machine learning library Scikit-learn in Python can be used. Furthermore, principal component analysis can be used to compress the dimensionality as needed. Moreover, the above-mentioned machine learning algorithm is merely one example and is not limited to it.

[0044] Figure 5 shows the construction of the neural network (NN) used as the inference model. Rectangles represent layers that perform data transformation, and rounded rectangles represent input and output data. D is the number of inspection items. The inference model in Figure 5 transforms data through four layers (Compile A1~Compile A3, Compile B) in the order of D-dimensional, 96-dimensional, 96-dimensional, 96-dimensional, and 1-dimensional.

[0045] Furthermore, as shown in Figure 5, Compile A1 to A3 include a fully connected layer "Linear" for fully connected processing, a "Kernel Regularizer" for regularization, and a "ReLU" layer for ReLU processing. Layer group Compile B includes a fully connected layer "Linear" and an "Adam" layer for optimization. The input units of the fully connected layers in layer group Compile A1 are equivalent to the input layer, the output units of layer group Compile B are equivalent to the output layer, and the units located between them are equivalent to the intermediate layers (hidden layers). The intermediate layers include a Dropout layer that controls a portion of the input values ​​to 0 to prevent overlearning, and a BN (Batch Normalization) layer that performs normalization in small batches during learning.

[0046] When XGBoost is used in machine learning via gradient boosting regression trees, the residual between the blood triglyceride level obtained from a blood test and the estimated blood triglyceride level is calculated, and the parameters of XGBoost (max_depth, subsample, colsample_bytree, learning_rate) are adjusted to minimize the mean squared error, thereby generating the estimation model. max_depth is the depth of the decision tree, subsample is the proportion of randomly selected samples from each tree, colsample_bytree is the proportion of randomly selected rows from each tree, and learning_rate represents the learning rate. Specifically, max_depth is adjusted within the range of 1 to 10, subsample within the range of 0.1 to 1.0, colsample_bytree within the range of 0.3 to 1.0, and learning_rate within the range of 0.1 to 0.7.

[0047] <Using a blood triglyceride estimation model to estimate blood triglycerides> Figure 6 is a flowchart showing the execution steps of the blood triglyceride estimation process. In step ST201, the first acquisition unit 31 of the blood triglyceride estimation device 30 acquires the user's attribute information from the terminal device 10. In step ST202, the second acquisition unit 32 of the blood triglyceride estimation device 30 acquires the user's non-invasive biological information. Then, the user's attribute information and non-invasive biological information are stored in the user data memory unit 33. Subsequently, in step ST203, the blood triglyceride estimation model stored in the estimation model memory unit 36 ​​is used to calculate the estimated blood triglyceride value by the estimation processing unit 37. In step ST204, the calculated estimated value of blood triglycerides is output to the estimated data memory unit 38 for memory. In step ST205, the estimated value of blood triglycerides is displayed on an external terminal such as the display device 39.

[0048] <Example (Estimation of Triglycerides in Blood)> The following describes an embodiment of the invention. However, the invention is not limited to the following embodiments. Attribute information includes any one or a combination of ID, name, age, and gender, and non-invasive biometric information includes any one or a combination of BMI, blood pressure, pulse wave data, electrocardiogram data, and bioimpedance. Height and weight, which serve as the basis for calculating BMI, are measured using a height meter and a weight scale, respectively, while blood pressure is measured using a blood pressure monitor. Furthermore, pulse wave data, electrocardiogram data, and bioimpedance are measured using the EstecBC-3 (dual-purpose system). Moreover, commercially available pulse wave meters, electrocardiographs, impedance measuring devices, and pulse oximeters can be used in combination to replace the EstecBC-3. Additionally, a pre-defined wearable terminal can also be used to acquire the aforementioned non-invasive biometric information.

[0049] The bioimpedance (conductance) was measured by passing a weak current between two of six electrodes: the feet, hands, and left and right foreheads. The voltage and current were set to 1.28 V and 200 μA, respectively, and the conductivity was measured every 1 second for 32 milliseconds. The current was passed between two of the six electrodes, and the following were measured: (1) the conductance of the anode / cathode (μS), (2) the conductance of the cathode / anode (μS), (3) the difference between the conductance measured by (1) and the conductance measured by (2) (∆SCRA-SCRC), and (4) the conductivity (μS / m). In addition, muscle mass, body fat mass, total water content, phase angle, and resistance value were measured, as well as the dielectric constant (μSi) when current was passed between the right hand and left hand and between the right forehead and left forehead.

[0050] Using the Estec BC-3, pulse wave data, electrocardiogram data, and bioimpedance data were measured for each subject for 2 minutes. During the measurement, an instrument with electrocardiogram, pulse wave meter, and pulse oximeter functions was installed on the index finger of the subject's left hand, and two electrodes were installed on the forehead. The subject then placed their hands and feet on the electrode plates while sitting in a chair.

[0051] <Learning Model 1> In learning model 1, as shown in Figure 4, machine learning is performed in parallel using linear regression, neural network (NN) and gradient boosting regression tree (XGBoost), and ridge learning (ridge regression) is then performed on the learning results. At this time, the following data are selected as attribute information and non-invasive biological data. (A) Attribute information・gender・age (B) Non-invasive biological data・BMI・blood pressure…mean arterial pressure, pulse pressure・pulse wave data…pulse, elasticity index, LVET (left ventricular ejection time), b / a, acceleration pulse wave aging index, DEI (diplastic elasticity index)・ECG data…heart rate, MxDMn ratio, LF / HF・biological impedance…body fat mass (kg), body fat mass (%), muscle mass, ESG2+4+15+17 (μS / m), ESG9+10 (μS / m), R (Ω), cardiac output, conductivity on the frontal pathway (9). Here, ESG9+10 refers to the average impedance measured at the location shown in Figure 2. [μS / m] is the unit of the measured average, and [%] is the value obtained by proportionally converting the measured average within the normally measurable range. Furthermore, the non-invasive biological data further includes pulse pressure / pulse. <Example 1> In Example 1, using the attribute information of a total of 712 subjects, (1) non-invasive biological information measured by a height and weight scale, a blood pressure monitor and EstecBC-3, and (3) a training dataset of blood triglycerides obtained by blood tests conducted on the same day as the non-invasive biological information measurement, a blood triglyceride estimation model was generated through machine learning of the above-mentioned learning model 1. Furthermore, in the machine learning, the blood triglyceride measurement values ​​obtained by blood tests were converted into logarithmic values. Then, the correlation coefficient between the logarithm of the blood triglyceride level and the logarithm of the blood triglyceride level obtained by blood test was calculated to determine the estimation accuracy of the blood triglyceride estimation model. The results showed that the average error between the estimated blood triglyceride level and the blood triglyceride level obtained by blood test was 12.4, with a standard deviation of 53.7 and a standard error of 4.03. The correlation coefficient between the logarithm of the blood triglyceride level obtained by blood test and the logarithm of the estimated blood triglyceride level obtained by the blood triglyceride estimation model was 0.63. Furthermore, for the results of no correlation test, the p-value was less than 0.001, indicating a correlation between the logarithm of the blood triglyceride level estimated by this estimation model and the blood triglyceride level obtained by blood test.

[0052] As described above, by using non-invasive biological data including BMI (body mass index), blood pressure, pulse wave data, electrocardiogram data and biological impedance, a blood triglyceride estimation model is generated through machine learning, which can estimate blood triglyceride levels without the need for blood tests.

[0053] Furthermore, even if the number of each data included in the non-invasive biological data is limited as shown below, it is still possible to determine the "risk of blood triglycerides", that is, whether the value of blood triglycerides is normalized.

[0054] <Generating a Risk Prediction Model for Blood Triglycerides via Machine Learning> According to the present invention, instead of calculating the estimated value of blood triglycerides, the calculation of the risk prediction value of blood triglycerides can be performed. Figure 7 is a flowchart showing the execution steps of generating a risk prediction model for blood triglycerides via machine learning. In step ST301, the learning processing unit 35 performs preprocessing of the input data (e.g., the training dataset mentioned above). Specifically, the learning processing unit 35 converts blood triglyceride levels below 150 mg / dL to 0 (no risk) and levels above 150 mg / dL to 1 (risk). Furthermore, when the number of people classified as 0 is disparate and unbalanced with the number of people classified as 1, the learning processing unit 35 can also apply SMOTE (Chawla, NV. et al. 2002) to the learning data to manually generate training samples. That is, labels indicating the presence or absence of blood triglyceride risk (e.g., 0 or 1 as mentioned above) can be added to the input data (training dataset). When the difference between the number of blood triglyceride risk (label: 1) and the number of blood triglyceride-free risk (label: 0) is greater than a predetermined value, sample data in the training dataset is increased (generated) to reduce the difference. In step ST302, in addition to (1) performing machine learning using the first neural network (NN) constructed as shown in FIG8, or (2) performing machine learning using the second neural network (NN) constructed as shown in FIG9, the learning processing unit 35 performs machine learning in parallel using logistic regression, gradient boosting regression tree, and random forest. When performing machine learning using gradient boosting regression tree, software libraries such as XGBoost, CatBoost, and LightBGM can be used. In step ST303, the learning processing unit 35 performs non-sampling learning on the learning results obtained by machine learning through logistic regression, the second neural network (NN), gradient boosting regression tree, and random forest. Furthermore, when only the first neural network (NN) is used for machine learning in step ST302, the non-sampling learning (voting) in step ST303 is not required.

[0055] Figure 8 shows the construction of the first neural network (NN) used as a risk estimation model for blood triglycerides. D represents the number of examination items. The first neural network (NN) is constructed through three layer groups (Compile A1, Compile A2, Compile B), transforming data in the order of D-dimensional, 96-dimensional, 64-dimensional, and 1-dimensional. Furthermore, Compile A1 and A2 contain fully connected layers "Linear" for fully connected processing, "Kernel Regularizer" for regularization, and "ReLU" layers for ReLU processing. Layer group Compile B contains fully connected layers "Linear" for fully connected processing and "Adam" layers for optimization. The input units of the fully connected layers in layer group Compile A1 are equivalent to the input layer, the output units of layer group Compile B are equivalent to the output layer, and the units located between them are equivalent to the intermediate layers (hidden layers). The intermediate layers include a Dropout layer that sets a portion of the input values ​​to 0 to prevent overlearning, and a Batch Normalization (BN) layer that performs normalization and passes the values ​​to the next layer.

[0056] Figure 9 shows the construction of the second neural network (NN) used as a risk estimation model for blood triglycerides. D represents the number of examination items. The construction of the second neural network (NN) involves data transformation through four layers (Compile A1, Compile A2, Compile A3, Compile B) in the order of D-dimensional, 64-dimensional, 64-dimensional, 64-dimensional, and 1-dimensional.

[0057] When performing machine learning using logistic regression, for example, the logistic regression provided by the open-source machine learning library Scikit-learn in Python can be used. Furthermore, if necessary, principal component analysis can be used to compress the dimensionality. Also, by comparing the risk of blood triglycerides obtained through blood tests with the inferred risk of blood triglycerides estimated through machine learning, the parameters of logistic regression (C, regularization method, max_iter, solber) are adjusted to maximize the F1 score. Here, C is a trade-off parameter determining the strength of regularization; a larger value indicates weaker regularization. The regularization method refers to either L1 regularization or L2 regularization, which is selected. max_iter is the maximum number of repetitions. solber is used to select the convergence method that minimizes the cross-entropy error (e.g., L-BFGS, Newton-CG, bilinear (liblinear), sag, and saga). Furthermore, in Example 2 below, the bilinear method was selected. Furthermore, the aforementioned machine learning algorithm is merely one example and is not limited to it. When performing machine learning using gradient boosting regression trees, software libraries such as XGBoost, CatBoost, and LightBGM can be used. Additionally, by comparing the blood triglyceride risk assessment value obtained through blood tests with the blood triglyceride risk estimation value derived through machine learning, the parameters of XGBoost (max_depth, subsample, colsample_bytree, learning_rate) are adjusted to maximize the F1 score. Here, max_depth is the depth of the decision tree, subsample is the proportion of randomly selected samples from each tree, colsample_bytree is the proportion of randomly selected rows from each tree, and learning_rate represents the learning rate. Specifically, max_depth is adjusted within the range of 1 to 10, subsample within the range of 0.3 to 0.7, colsample_bytree within the range of 0.3 to 1.0, and learning_rate within the range of 0.1 to 0.9. When performing machine learning using random forests, for example, the Random Forest Classifier provided by the open-source machine learning library Scikit-learn in Python can be used. Here, the risk assessment value of blood triglycerides obtained through blood tests is compared with the inferred risk value of blood triglycerides estimated through machine learning, and the parameters of the random forest classifier (max_depth, max_features, n_estimators) are adjusted to maximize the F1 score.`max_depth` is the depth of the decision tree, and `max_features` specifies the randomness of each tree. `n_estimators` is the number of decision trees to be created. Specifically, `max_depth` can be set to None or adjusted within the range of 1 to 7, `subsample` can be adjusted within the range of 0.3 to 0.7, `max_features` should be the best among "auto", "sqrt", and "log", and `n_estimator` can be adjusted within the range of 10 to 1000.

[0058] <Using a blood triglyceride risk estimation model to estimate blood triglyceride risk> As shown in Figure 10, in step ST401, the first acquisition unit 31 of the blood triglyceride estimation device 30 acquires the user's attribute information from the terminal device 10. In step ST402, the second acquisition unit 32 of the blood triglyceride estimation device 30 acquires the user's non-invasive biological information. Then, the user's attribute information and non-invasive biological information are stored in the user data memory unit 33. Subsequently, in step ST403, using the blood triglyceride risk estimation model stored in the estimation model memory unit 36, the estimation processing unit 37 calculates the blood triglyceride risk estimation value, i.e., the probability of belonging to level 0 (no risk) or level 1 (risky). In step ST404, the calculated blood triglyceride risk estimate is output to the estimate data memory unit 38 for memory storage. In step ST405, the blood triglyceride risk estimate is displayed on an external terminal such as the display device 39.

[0059] <Example (Risk Presumption of Blood Triglycerides)> The following describes an example of risk presumption of blood triglycerides. However, the form of risk presumption of blood triglycerides in this invention is not limited to the following examples. Attribute information includes any one or a combination of ID, name, age, and gender, and non-invasive biometric information includes any one or a combination of BMI, blood pressure, pulse wave data, electrocardiogram data, and bioimpedance. Height and weight, which serve as the basis for calculating BMI, are measured using a height meter and a weight scale, respectively, while blood pressure is measured using a blood pressure monitor. Furthermore, pulse wave data, electrocardiogram data, and bioimpedance are measured using the EstecBC-3 (dual-purpose system). Alternatively, commercially available pulse wave meters, electrocardiographs, impedance measuring devices, and pulse oximeters can be used in combination to replace the EstecBC-3. Also, a pre-defined wearable terminal can be used to acquire the aforementioned non-invasive biometric information. Bioimpedance (conductance) was measured by passing a weak current between two of six electrodes located on the feet, hands, and left and right foreheads. The voltage and current were set to 1.28 V and 200 μA, respectively, and the conductivity was measured every 1 second for 32 milliseconds. Current was passed between two of the six electrodes, and the following were measured: (1) the conductance at the anode / cathode (μS), (2) the conductance at the cathode / anode (μS), (3) the difference between the conductance measured in (1) and the conductance measured in (2) (∆SCRA-SCRC), and (4) the conductivity (μS / m). Muscle mass, body fat mass, total water content, phase angle, and resistance were also measured. The dielectric constant (μSi) was also measured between the right and left hands and between the right and left foreheads when current was applied.

[0060] Using the Estec BC-3, pulse wave data, electrocardiogram data, and bioimpedance data were measured for each subject for 2 minutes. During the measurement, an instrument with electrocardiogram, pulse wave meter, and pulse oximeter functions was installed on the index finger of the subject's left hand, and two electrodes were installed on the forehead. The subject then placed their hands and feet on the electrode plates while sitting in a chair.

[0061] <Learning Model 2> In Learning Model 2, machine learning is performed using the first neural network (NN) constructed as shown in Figure 8. At this time, the following data are selected as non-invasive biological data. (A) Attribute information・Age (B) Non-invasive biological data・BMI・Blood pressure…systolic blood pressure, pulse pressure・pulse wave data…pulse・bioresistance…fat-free body weight (kg), total water content (%), 13 left foot - 14 right foot / SCR A, ESG 6+13+19 (%).

[0062] <Learning Model 3> In Learning Model 3, in parallel with machine learning via the second neural network shown in Figure 9, machine learning is performed using logistic regression, gradient boosting, and random forest, and non-sampling learning is performed on the learning results using a voting method, thereby creating a risk estimation model for triglycerides in bleeding. At this time, the data shown below are selected as non-invasive biological data. Furthermore, unlike Learning Model 2 above, Learning Model 3 does not require subject attribute information. (B) Non-invasive biological data・BMI・Blood pressure…systolic blood pressure, diastolic blood pressure・Pulse wave data…elasticity index, LVET, pulse・Bioimpedance…body fat mass (kg), body fat percentage (%), muscle mass, total water content (%), R (Ω), cardiac output, 13 left foot - 14 right foot / SCR A, EGS 9 + 10 (μS / m).

[0063] <Example 2> In Example 2, using (1) attribute information of 321 subjects, (2) non-invasive biometric information measured by a height and weight scale, a blood pressure monitor, and an Estec BC-3, and (3) a training dataset of blood triglyceride levels obtained from blood tests conducted on the same day as the non-invasive biometric information measurements, a blood triglyceride risk prediction model was generated through machine learning of the learning model 2 described above. Then, the prediction accuracy of the blood triglyceride risk prediction model was evaluated using the ROC_AUC curve. The result showed that the ROC_AUC was 0.80, indicating that it was classified as extremely good. The ROC_AUC curve of the prediction result of Example 2 is shown in Figure 11.

[0064] <Example 3> In Example 3, using (1) attribute information of a total of 712 subjects, (2) non-invasive biometric information measured by a height and weight scale, a blood pressure monitor, and an Estec BC-3, and (3) a training dataset of blood triglyceride levels obtained from blood tests conducted on the same day as the non-invasive biometric information measurements, a blood triglyceride risk prediction model was generated through machine learning of the learning model 3 described above. Then, the prediction accuracy of the blood triglyceride risk prediction model was evaluated using the ROC_AUC curve. The result showed that the ROC_AUC was 0.77, which exceeded 0.7, indicating a good classification. The ROC_AUC curve of the prediction result of Example 3 is shown in Figure 12.

[0065] (Variation Example) In the above embodiment, an example of using learning model 2 or learning model 3 to estimate the risk of blood triglycerides has been described. However, several learning models can also be used to estimate the risk of blood triglycerides. In this way, the estimation can be performed with higher accuracy compared to using one learning model to estimate the risk of blood triglycerides.

[0066] Furthermore, in the above embodiments and examples, there have been instances where the estimation accuracy is improved when BMI is used as the dataset for the learning model. Therefore, a function to estimate the BMI can also be provided in the blood triglyceride estimation device. BMI is generally not obtainable through wearable terminals, but is calculated based on the height and weight input by the user. By estimating the BMI, the aforementioned risk of blood triglycerides can be obtained simply by acquiring biological information, thus improving convenience for the user. The method of BMI estimation is not particularly limited, but for example, it is known that BMI is correlated with the tilt of the user's abdomen (at a predetermined position). Therefore, for example, a predetermined accelerometer can be installed on the user's abdomen (or a wristband-type wearable terminal with an accelerometer can be placed against the abdomen), and the tilt of the abdomen can be calculated based on the data output from the accelerometer to estimate the BMI.

[0067] Similarly, as described above, a wristband-type wearable terminal equipped with a pulse wave sensor or a blood oxygen concentration sensor can also be used to estimate pulse wave data.

[0068] Similarly, as described above, a wristband-type wearable terminal or the like can also be used to estimate blood pressure. Since it is known that the velocity of the pulse wave transmitted through the arteries by the heartbeat is correlated with blood pressure, a pre-existing sensor that measures the velocity of the pulse wave transmitted through the arteries by the heartbeat can also be used to estimate blood pressure.

[0069] Similarly, as described above, wearable devices such as wristbands can also be used to estimate electrocardiogram (ECG) data. For example, ECG data can be estimated based on data obtained from electrodes disposed on the side of the wristband-type wearable device opposite to the display surface and electrodes disposed on the display surface side. Specifically, the wrist of the hand wearing the wristband-type wearable device (e.g., the left hand) can be connected to the electrodes disposed on the opposite side, and the fingertips of the hand opposite to the hand wearing the wristband-type wearable device (e.g., the right hand) can be connected to the electrodes disposed on the display surface side, and ECG data can be estimated based on the data obtained therefrom.

[0070] Furthermore, similarly to the above, bioimpedance can also be estimated using wearable devices such as wristbands equipped with various electrodes. For example, bioimpedance can be estimated using wearable devices such as wristbands based on bio-information obtained from the chest and wrist.

[0071] Furthermore, biological information obtainable through a wearable terminal and biological information of the aforementioned presumed object (at least one of BMI, blood pressure, pulse wave data, electrocardiogram data, and biological impedance) can be used as teaching materials, and a classifier generated using various machine learning algorithms can be used to presumably determine the biological information of the aforementioned presumed object. Furthermore, in this case, the first acquisition unit can also acquire the presumed biological information.

[0072] (Other) Furthermore, for example, the above series of processes can be executed by hardware or software. In other words, the above functional configuration is merely illustrative and not particularly limited. That is, as long as the information processing system has the function of executing the above series of processes as a whole, it is sufficient; the specific functional module used to implement the function is not particularly limited to the above example. Furthermore, the location of the functional module is not particularly limited to that shown in Figure 4, but can be arbitrarily set. For example, the functional module of the server can be ported to other terminals or devices, etc. Conversely, the functional module of other terminals or devices can be ported to the server, etc. Furthermore, a functional module can be composed of a single hardware unit, a single software unit, or a combination thereof.

[0073] In cases where a series of processes are performed by software, the program constituting the software is installed from the network or recording media onto a computer, etc. The computer may be a computer assembled in dedicated hardware. Alternatively, the computer may be a computer that can perform various functions by installing various programs, such as a server and other general-purpose smartphones or personal computers.

[0074] The recording medium containing such a program can be either a portable medium (not shown) that is separately configured from the device body for providing the program to users, or a recording medium that is provided to users in a state of pre-assembled in the device body. The program can be transmitted via a network, therefore the recording medium can also be a computer that is connected to or can be connected to a network.

[0075] Furthermore, in this specification, the steps of a program recorded on a recording medium naturally include processing performed in sequence and in a time series, as well as processing that is not necessarily performed in a time series but is performed in parallel or individually. Also, in this specification, the term "system" means an overall device that includes several devices or means. [Simplified Explanation of the Diagram]

[0020] Figure 1 is a block diagram showing the general structure of the blood triglyceride estimation system. Figure 2 is a diagram illustrating an Electron Scan Gram (ESG). Figure 3 is a hardware diagram of the blood triglyceride estimation device. Figure 4 is a flowchart showing the execution steps of generating a blood triglyceride estimation model by machine learning. Figure 5 is the hierarchical structure of the neural network used for blood triglyceride estimation. Figure 6 is a flowchart showing the execution steps of blood triglyceride estimation processing. Figure 7 is a flowchart showing the execution steps of generating a blood triglyceride risk estimation model by machine learning. Figure 8 is the hierarchical structure of the first neural network used for blood triglyceride risk estimation. Figure 9 is the hierarchical structure of the second neural network used for blood triglyceride risk estimation. Figure 10 is a flowchart showing the execution steps of blood triglyceride risk estimation processing. Figure 11 shows the ROC-AUC curve of the estimated results in Example 2. Figure 12 shows the ROC-AUC curve of the estimated results in Example 3.

Claims

1. A device for estimating blood triglycerides, characterized by comprising: an information acquisition unit that acquires non-invasive biological information of a predetermined user, including at least one of blood pressure and electrocardiogram data; an estimation model memory unit that stores a blood triglyceride estimation model; and an estimation processing unit that uses the blood triglyceride estimation model to calculate the estimated blood triglyceride value of the predetermined user based on the user's attribute information and non-invasive biological information.

2. The apparatus for estimating blood triglycerides as described in claim 1, wherein, The aforementioned attribute information includes either age and gender, or a combination thereof. The aforementioned non-invasive biological information includes either BMI, blood pressure, pulse wave data, electrocardiogram data, and biological impedance, or a combination thereof.

3. The blood triglyceride estimation device as claimed in claim 1 or 2 further comprises: a training data memory unit that stores a training data set; and a learning processing unit that generates the blood triglyceride estimation model based on the training data set by means of machine learning.

4. The apparatus for estimating blood triglycerides as described in claim 3, wherein, The training dataset includes the subject's attribute information, non-invasive biological information, and blood triglyceride levels measured from blood samples.

5. The apparatus for estimating blood triglycerides as described in claim 4, wherein, The correlation coefficient between the logarithm of the above-mentioned estimated values ​​of blood triglycerides and the logarithm of the above-mentioned measured values ​​of blood triglycerides is greater than 0.

6.

6. The apparatus for estimating blood triglycerides as described in claim 3, wherein, The training dataset includes non-invasive biological information of the subjects and blood triglyceride levels measured from their blood. The aforementioned presumption processing unit replaces the aforementioned blood triglyceride presumption values ​​with a presumption risk value for blood triglycerides.

7. The apparatus for estimating blood triglycerides as described in claim 6, wherein, The aforementioned learning processing unit adds labels to the aforementioned training dataset indicating the presence or absence of the aforementioned risk of blood triglyceride levels based on blood triglyceride measurement values. When the difference between the number of individuals with the aforementioned risk of blood triglyceride levels and the number of individuals without the aforementioned risk of blood triglyceride levels exceeds a predetermined value, the unit increases the sample data in the aforementioned training dataset to reduce the aforementioned difference.

8. The apparatus for estimating blood triglycerides as described in claim 3, wherein, The learning processing unit generates a first blood triglyceride risk estimation model and a second blood triglyceride risk estimation model through machine learning based on different types of training datasets. The estimation processing unit uses the first blood triglyceride risk estimation model and the second blood triglyceride risk estimation model to calculate the blood triglyceride risk estimation value of the given user.

9. The blood triglyceride estimation device of claim 1 or 2 further includes a biological information estimation unit that estimates at least one of the biological information included in the above-mentioned biological information, such as BMI, blood pressure, pulse wave data, electrocardiogram data and biological impedance. The information acquisition unit acquires the biological information estimated by the biological information estimation unit as the biological information of the predetermined user.

10. A non-invasive blood triglyceride estimation system, characterized in that: it comprises the blood triglyceride estimation device of any one of claims 1 to 9, and a biological information measuring device for measuring non-invasive biological information.

11. A method for estimating blood triglycerides, comprising the following steps: memorizing a training dataset containing non-invasive biological information of at least one of the subjects, including blood pressure and electrocardiogram data, and blood triglyceride measurement values ​​obtained from blood; generating a blood triglyceride estimation model by machine learning based on the training dataset; and calculating the estimated blood triglyceride value of the predetermined user based on the blood triglyceride estimation model and the non-invasive biological information of the predetermined user.

12. A program that enables a computer to perform the following steps: memorizing a training dataset containing non-invasive biological information of at least one of the subjects, including blood pressure and electrocardiogram data, and blood triglyceride measurement values ​​obtained from blood; generating a blood triglyceride estimation model by machine learning based on the training dataset; and calculating the blood triglyceride estimation value of the predetermined user based on attribute information and non-invasive biological information of the predetermined user using the blood triglyceride estimation model.

Images