TWI500411B - Assessment system and method for pulse and constitution health risks - Google Patents

Description

Pulse wave and physical health risk assessment system and method

The disclosure relates to a seismic and constitutional health risk assessment system and method.

Social changes have gradually changed the spectrum of human diseases, from early infectious diseases to heart disease, diabetes, stroke, high blood pressure, cancer and other chronic diseases. According to the statistics of relevant data, the causes of death such as old age and chronic diseases have gradually become the main cause of affecting the average life expectancy in the future. Early techniques used only the analysis of pulse data to determine health status. As shown in the example of the first figure, the horizontal axis represents the pulse wave parameter, the vertical axis represents the disease risk, and R1 to R4 represent the disease risk curves of the four ethnic groups. This is easy because of the limited pulse wave data, resulting in an overestimation of the state of health, that is, underestimating the risk of high-risk groups, and losing the opportunity to grasp the state of health in a timely manner.

The method of evaluation combined with pulse diagnosis and physical information is in line with the syndrome of TCM medical theory. Pulse disease is the main basis for clinical syndrome differentiation of traditional Chinese medicine. The syndrome here refers to the abnormal sensation and morphological changes of the patient collected through the four clinics, such as the constitution; the pulse refers to the pulse. The idea of complication of pulse syndrome has been clearly stated in the literature of traditional Chinese medicine. For example, there is a Chinese medical literature that mentions that “the pulse is static and the savvy, the five colors are observed, the five internal organs are observed, there are more than enough, the six squats are strong and weak, and the shape is ups and downs. "Wu, the difference between death and death", and "joining and doing, can be for work." Later, there were also Chinese medical literatures that revealed the veins. The principle of syndrome differentiation is applied to the clinic. For example, in the disclosure of the Chinese medical literature, it emphasizes "viewing its pulse syndrome, knowing what the crime is, and treating it with the syndrome", and the title of "and certain disease and syndrome". The disclosure of these documents not only pays attention to the analysis of the internal relationship between the pulse and the disease, but also pays more attention to the exploration of the pathogenesis when the pulse disease is opposite, so as to more accurately judge the true cause of the yin and yang qi and blood stasis.

Pulse diagnosis has the power to rapidly differentially diagnose and assess conditions. It can be used to measure the number, position, shape, potential, and law of the radial artery, and to understand the ups and downs of the internal organs. Such non-invasive diagnostics are effective and harmless and provide a wealth of information. Traditional Chinese medicine pulmonology has complete diagnosis and treatment of blood flow pulsation in human physiology. One of the theories of traditional Chinese medicine is to observe the coordination state of each organ, and the pulsation is the fluctuation of blood flow, which represents the relationship between each organ and the vibration, so the blood flow pulsation is analyzed to evaluate the physiological condition. It can be seen from the above that the change of blood flow pulsation is an important message for the physiological condition of the reaction.

In the prior art related to the generation and interpretation of this message, for example, there is a document proving that the structure of blood vessels has a significant influence on the waveform of blood pressure waves; there is also a literature indicating that the blood vessels of some organs and tissues have the most frequency selectivity, that is, each organ and The tissue only allows blood pressure waves of some specific frequencies to flow, and the high-frequency blood pressure waves have the same resistance for each organ. Therefore, the interaction of various organ tissues can be seen, and the TCM pulsology is to observe this phenomenon to judge the physiological state of the human body. In every part of the whole blood vessel system, including the heart, arteries, vascular plexus, veins and related nerves and endocrine systems, its mechanical properties, electrical properties and geometric distribution characteristics are interlocking and precise. Complex control, so blood pressure and blood flow pulsation provide a rich and varied physiological message, which is an indispensable parameter for studying the characteristics of the cardiovascular system.

From the arterial side to the venous side, the maximum pressure drop of the blood pressure wave occurs in the vascular plexus composed of the arterioles and microvessels. Therefore, the characteristics of the vascular plexus are factors that can not be ignored in the blood flow pulsation, and the organ tissues in the human body contain Extremely complex vascular plexus, and the amount of blood required, the structure of the vascular plexus and the relaxation of the arterioles are also different. The vascular plexus dominates the functions of various organ tissues, each with different characteristics, so the characteristics of each organ tissue will be expressed in some way on the blood flow pulsation waveform. Therefore, the analysis and discussion of blood flow pulsation changes can show the corresponding physiological organ tissue state.

There are many related technologies for the analysis and discussion of existing blood flow pulsation changes. For example, one technique is to provide a blood pressure monitor that simultaneously outputs blood pressure measurements and the degree of risk of developing a disease (eg, cardiovascular disease). This technique uses a single piece of blood pressure data and information on the level of cholesterol to be calculated. This information on cholesterol levels is obtained by first taking a blood test. One technique discloses a cardiovascular function evaluation device that simultaneously measures ECG signals and vascular pulse signals, and calculates a Stiffness Index (SI) and a Reflex Index (RI). And use the synchronous measured ECG signal to calculate Pulse Wave Velocity (PWV); this technology can provide users with complete information as a reference for assessing heart and blood vessel health.

There is a technology for providing an analytical system and method for pulse diagnosis of a Chinese medicine. The technique uses a pulse wave collecting device to collect and generate blood pressure signals and ECG signals, and a signal processing unit to construct blood pressure waveform parameters and arterial pressure variability (Arterial Pressure Variability, APV) and Heart Rate Variability (HRV) spectrum analysis parameters. And based on these parameters to define the quantitative indicators of the components of TCM pulse. One technique is to provide a pulse wave analysis processing device. This technique compares the pulse signals acquired at different locations and at different times, and cross-compares these pulse signal feature points to improve the accuracy of the pulse diagnosis signal analysis.

There are some related technologies for the analysis and discussion of the above-mentioned and existing blood flow pulsation changes, and some technologies are invasive and cannot obtain data immediately. Some techniques do not reveal how to perform physiological function assessment of physiological parameters. Some technologies do not fully disclose how the signal parameters represent the elements of TCM pulse and form quantitative indicators. Some technologies do not combine personal physique parameters to fully assess personal health status. Therefore, there is a need to design a non-invasive health assessment tool. This health assessment tool combines arterial waveform data with personal physique classification information to provide instant information, allowing users to instantly understand their physical condition and enhance their awareness of self-health management. It can be used for daily life to prevent disease or delay onset.

The disclosed embodiments provide a pulse wave and physique health risk assessment system and method.

An embodiment of the present disclosure is directed to a pulse wave and physique health risk assessment system. The system can include a pulse sensing device, a constitution identification device, and a risk assessment device. The pulse wave sensing device senses a plurality of pulse signals of at least one user. The physique recognition device recognizes the physique corresponding to the at least one user. The risk assessment device estimates at least one disease corresponding to the at least one user based on a set of pulse wave characteristic parameters obtained by the plurality of pulse wave signals and a set of body mass classifications obtained by the at least one user Risk indicator information.

Another embodiment of the present disclosure is directed to a pulse wave and physique health risk assessment method. The method includes: obtaining, by a pulse wave sensing device, a plurality of pulse signals of the at least one user, obtaining a set of pulse wave characteristic parameters of the at least one user, and identifying the corresponding by using the integrated quality identification device At least one user of the at least one user; defining, by the risk assessment device, the corresponding at least one user group according to the at least one entity state of the at least one user, and according to the corresponding at least one user group Defining respective corresponding disease risk criteria or risk models; and estimating, via the risk assessment device, the at least one user's set of pulse wave characteristic parameters and their corresponding disease risk criteria or risk models At least one user's at least one disease risk level.

The above and other advantages of the present invention will be described in detail below with reference to the following drawings, detailed description of the embodiments, and claims.

The pulse wave and body health risk assessment technology of the embodiment of the present disclosure analyzes the user's pulse wave characteristics by extracting user pulse wave information, and identifies the user's physical fitness category, and then defines different according to different physical characteristics. Pulse-risk model or risk criteria. The user's disease risk level is then defined according to the above risk model or risk criteria, such as high risk, medium risk, and low risk. In this way, the user can output relevant information such as the disease risk level or the disease risk index, and provide important reference information for personal health management.

In the present disclosure, physique refers to an individual's traits, including (physical) structural and functional traits, personality, adaptability to environmental changes, and resistance to disease. Physical fitness is a relatively stable nature, partly determined by genes, and partly by acquired genes.

The disclosed embodiment utilizes an indicator risk assessment model, such as a logistic regression model, to treat the two sets of health risk parameters as inputs, and the health risk assessment values as outputs to derive health risk assessment information that is absolutely relevant to the constitution. Among the two groups of health risk parameters, one set of health risk parameters is a set of pulse wave characteristic parameters, and the other set of health risk parameters is a K type of physique that can be distinguished by the classification of constitutional identification.

In view of the above, the second figure is a pulse wave and physical health risk assessment system according to an embodiment of the present disclosure. Referring to the second figure, the health risk assessment system 200 can include a pulse wave sensing device 210, an integrated quality identification device 220, And a risk assessment device 230. The pulse wave sensing device 210 senses a plurality of pulse signals of at least one user. The physique recognition device 220 identifies at least one genre classification 222 corresponding to the at least one user. The risk assessment device 230 evaluates at least one disease corresponding to the at least one user according to a set of pulse wave characteristic parameters obtained by the plurality of pulse wave signals and a set of body mass classifications obtained by the at least one user The risk information 232 may be, for example, a disease risk index, a disease risk level, a disease risk level, or the like.

The pulse wave sensing device 210 can sense a plurality of pulse signals of the user by using a pressure type or a microphone type; the plurality of pulse wave signals extracted by the pulse wave can be parsed by using at least one numerical method by, for example, a pulse wave analyzer. , including time domain or frequency domain pulse wave analysis, to define one or more features of the plurality of pulse signals. A set of pulse wave characteristic parameters 212 is calculated. Numerical methods may include, but are not limited to, time domain or frequency domain pulse wave analysis to define one or more features of a plurality of pulse signals. The physique recognition device 220 can identify at least one of the questionnaires and at least one of the physiological signal acquisition methods to identify the physique classification corresponding to the user. The disease risk information evaluated by the risk assessment device 230 may include, but is not limited to, disease risk information of at least one circulatory system (eg, cardiovascular). The risk assessment device 230 may use the user's physique classification to establish a disease risk criterion for different users, or may use the set of pulse wave characteristic parameters and define the disease risk of the different users according to different users' disease risk criteria.

A variety of pulse wave analysis methods can be used to obtain a set of pulse wave characteristic parameters. The classification of constitutional identification can also be obtained through the physique identification tool. The following are explained separately Several examples of obtaining two sets of health risk parameters. The pulse wave analysis method measures, for example, a blood vessel waveform change of the radial artery through a TCM pulse measuring device, and extracts at least a first main wave to a third main wave in a single periodic wave through an algorithm. Using the wave height, wave width and spectral energy distribution information of the main wave, it can be used as the first set of parameters for health risk assessment. As shown in the third figure, the pulse signal map of a typical age group, where the horizontal axis represents time, The vertical axis represents the pulse signal height h, and the first set of parameters for health risk assessment is a set of pulse wave feature points including h1 for main wave amplitude, h2 for pre-pulse amplitude, and h3 for re-pulse amplitude. , t1 represents the acute ejection phase, t2 represents the systolic phase, and t represents the pulse wave period. The fourth to fourth C charts are pulse wave signals of three age groups, respectively, wherein the three age groups are 25-age, 47-age, and 80-age.

In the pulse analysis, a time domain analysis method can be used to take the first to fourth derivative of the pulse signal curve, and the pulse wave characteristic value parameters are extracted (for example: main wave amplitude, pre-pulse amplitude, pulse wave contraction) period). Observe the one to four derivatives of the pulse wave: f ' ( t ), f " ( t ), f ''' ( t ), f "" ( t ), relative to the position of the zero point, as shown in the fifth figure The feature points and their eigenvalues can be found on the original waveform (including h1~h3, t1, t2, t in the third figure). Several pulse wave eigenvalues with discriminative ability can also be extracted from the 。. In the fifth figure, the algorithm for calculating the pulse wave characteristic value is explained as follows.

1. Using the second derivative, find the highest point of the waveform (maximum value is the systolic pressure, minimum value is the diastolic pressure).

2. Observe the fourth derivative of the position of the maximum value. If the slope of the fourth derivative is negative, then the maximum value is found to be early systolic.

3. From the fourth derivative, look for the third position of the late systolic position through the zero point (the line segment is from bottom to top), then the late systole can be found.

4. Find the time position of the heavy wave front wave, and then corresponding to the original waveform, you can find the measurement value needed to calculate the pulse wave characteristic value.

Another method of pulse wave analysis is to use Principal Components Analysis (PCA) technology to compare the data of the main features and extract the most important k feature information from multiple feature information. For its characteristic basis. The method used in principal component analysis is to calculate the covariate matrix. The eigenvalues and eigenvectors of the covariate matrix are calculated. After the eigenvalues and the corresponding eigenvectors are sorted, the functions of the main k features are taken as the main features, which can effectively reduce the dimension of the data.

In addition to the aforementioned fourth-order differential method and PCA method, the analysis of the pulse wave can also pass the Hilbert-Huang transform (HHT) method, and the analysis process can include finding the local maximum value and its envelope, finding the local minimum value and its envelope, The mean envelope is obtained from the envelope of the maximum value and the envelope of the minimum value, and the difference between the original signal and the mean envelope is used to obtain the component. After repeated operations and appropriate convergence, the pulse analysis results are obtained and can be used as a set of input information. The analysis result is the fifth to eighth modal functions C5 to C8 shown in the sixth figure, wherein the local maximum value of the sixth modal function C6 can be used to find the characteristic points h1~h3 of the original pulse signal. , t1, t2, t are used to capture the aforementioned pulse wave characteristics.

As mentioned earlier, it is permeable to the physique identification tool (for example, Wang's physique Class can distinguish human physique into nine physiques, including peace, yang deficiency, yin deficiency, qi deficiency, phlegm, dampness, blood stasis, qi stagnation, and special sputum. Classified information as a second set of parameters for health risk assessment. Taking Wang's constitution as an example, the rules for determining the constitution of the constitution are as follows. (1) Judgment method: Answer all the questions in the "Traditional Chinese Medicine Classification and Judgment Table". Each question is given 5 and the score, the original score and the conversion score are calculated, and the constitution type is determined according to the standard. The original score = the score of each item and the conversion score = [(original score - number of entries) / (number of entries × 4)] × 100; (2) Judging criteria: the peace value is normal constitution, and the other 8 constitutions are biased Physical fitness, the criteria for judgment are shown in Figure 7.

After having two sets of health risk parameters (ie, a set of pulse wave characteristics and a set of physique classifications), an indicator risk assessment model, such as the logistic function logit P, is used to enter the two sets of health risk parameters as health risk assessment values. For the output, it is possible to derive health risk assessment information that is absolutely relevant to the constitution. It is assumed that the classification of constitutional identification can be divided into K physiques. The risk assessment model of this indicator is described below.

Among them, β ₁ , β ₂₁ , β ₂₂ , β ₂₂ ,. . . , β _2k are all parameters; (pulse wave characteristics), (physique) ₁ ,. . . (Physique) _{k is} a variable; (pulse wave feature) can be a one-dimensional or multi-dimensional vector, and the parameter β ₁ corresponds accordingly, that is, the dimension of the (pulse wave feature) is equal to the dimension of β ₁ .

The eighth figure is an example of an index risk assessment model and an example of a logistic regression model as an index risk assessment model to derive health risk assessment information that is absolutely relevant to physical fitness, according to an embodiment of the present disclosure. Where i=1,2,3,...,n,n is the number of samples in the database, and P _i is the probability that the i-th sample will get the disease. According to a most approximate method, the parameters β ₁ , β ₂₁ , β ₂₂ , β ₂₂ , can be obtained. . . The estimated value of β _2k is such that the following approximate function is the maximum value.

Approximate function , j is the number of diseases obtained from the database

The ninth figure is a schematic diagram showing an example of defining different pulse wave-risk evaluation models under different constitutions according to an embodiment of the present disclosure. As shown in the ninth figure, in the case of the same pulse signal feature, depending on the different physical characteristics of the user, it may correspond to different disease risk levels, that is, the case of the same pulse signal feature in the disclosed embodiment. In the following, different pulse-risk models can be defined according to different physical characteristics. In the actual system design, the disclosed embodiment may also be as shown in the tenth figure. When the user's disease risk level is identified according to the pulse signal, the user may be given different risk levels according to different physical characteristics. For example, high, medium and low risk level values. Accordingly, the evaluation rule according to the embodiment of the present disclosure is as shown in FIG. 11 , that is, for each physique _j , the evaluation index = β ₁ × pulse wave characteristic value + β _2j , j = 1, 2, 3, ...,k.

From the database, the evaluation index value of each sample can be obtained through calculation. Taking the physique as a group, there are evaluation index values of the samples in the small to large sorting group, and the 25% percentile and 75% percentile are found as the tangent point of the three-point method, as in the example of the twelfth figure. Shown.

In the above, the thirteenth diagram illustrates a pulse wave and physical health risk assessment method according to an embodiment of the present disclosure. Referring to FIG. 13 , the pulse wave and body health risk assessment method first obtains a set of pulse wave characteristic parameters of the at least one user by using a plurality of pulse signals of at least one user sensed by a pulse wave sensing device. (Step 1310); and identifying, by the integrated quality identification device, at least one integrated state corresponding to the at least one user (step 1320); and utilizing a risk assessment device 230, according to at least one corresponding use by the at least one user The population groups define their respective disease risk criteria or disease risk models (step 1330). Through the risk assessment device 230, the method estimates the at least one disease risk level of the at least one user by using the set of pulse wave characteristic parameters of the at least one user and the corresponding disease risk standard or disease risk model ( Step 1340). The information on the degree of risk of the at least one disease is, for example, but not limited to, one or more of the foregoing information of a disease risk index, a disease risk level, and a disease risk level.

In step 1310, at least one numerical method can be used to calculate The set of pulse wave characteristic parameters may be included, and the numerical method may include time domain or frequency domain pulse wave analysis to define one or more features of the plurality of pulse wave signals. In step 1320, at least one of the questionnaires and the at least one physiological signal acquisition method may be used to identify at least one of the at least one user's at least one quality state. The physical state refers to the state of the individual's traits, including the physical and functional traits, personality, adaptability to environmental changes, and resistance to disease. In step 1330, the corresponding at least one user group may be defined according to the at least one entity state of the at least one user. In step 1340, different disease risk criteria or disease risk models may be defined for user populations of different physiques corresponding to the same pulse wave parameters. The disease risk standard or disease risk model can be an empirical level and one of the aforementioned ones.

An example of a pulse wave and body health risk assessment meter technology that implements the disclosed embodiments can be as shown in FIG. 14 . A health risk evaluator 1400 can include a user detection device 1410, an integrated quality classifier 1420, and a pulse. Wave analyzer 1430, and a risk assessment device 230. When the health risk evaluator 1400 is in operation, the user detection device 1410 captures the pulse wave and body information of a user; and then the body classifier 1420 distinguishes the user's physique. The risk assessment device 230 defines different pulse-risk models (for example, the ninth map) according to different constitutions or can use the three-point method (for example, the twelfth map) to find the tangent points to define different pulse-risk levels ( For example, the tenth figure); the above-mentioned user's physique classification information, and the pulse wave characteristic of the user analyzed by the pulse wave analyzer 1430, can be according to the above pulse The wave-risk model or pulse-risk level defines the user's disease risk level; the user's disease risk level information can be output via an output device. The disease risk criteria or disease risk model is for example from the one-piece database 1440.

Examples of the indicator risk assessment model In addition to the logistic function, other examples can be used, such as the algorithm of the neural network to calculate the health risk assessment information. The following is an example of a neural network-like algorithm that explains how to calculate health risk assessment information.

The architecture of the Backpropagation Neural Network (BPN) is Multilayer Perceptron (MLP). The commonly used learning algorithm is Error Back Propagation (EBP). The inverted transfer neural network is a combination of MLP plus EBP. The inverted transfer neural network includes an input layer, a hidden layer, and an output layer, and the actually active neurons have a hidden layer and an output layer. The number of neurons in the input layer and the output layer depends on the form of the problem. The number of neurons in the hidden layer is determined by trial and error. In the network, the values of the weights are used to link the neurons between the layers. The input layer is directly transmitted to the hidden layer. After weighted accumulation, an output value can be obtained through the activation function conversion. After the weighted accumulation, the output value is converted by the activation function and then transmitted to the output layer.

The inverse transfer-like neural network model utilizes the steepest slope method (Gradient Steepest Descent Method) minimizes the error function. The learning process is usually a learning round (Learning Epoch). A network can train the paradigm to learn repeatedly until the learning of the network reaches convergence. In the process of learning, each input data has a corresponding expected output value to supervise the learning of the network. The goal of learning is to adjust the connection weights between processing units to reduce the gap between the network inference output value and the expected value. The process of learning usually requires multiple cycles and a long time to get better results.

In the architecture of the inverted neural network of the fifteenth figure, the input layer is 1510 for representing the input variable x _{i of the} network, and x _i is the output value of the i-th neuron, and the number of processing units is Depending on the problem. Therefore, according to the disclosed embodiment, the pulse characteristic parameter and the k physique can be used as the input layer input variables, that is, the variable x ₁ is the pulse wave characteristic parameter, the variable x ₂ is the physique 1, and the variable x ₃ is the physique 2 ..., the variable x _k+1 is the constitution k; its hidden layer 1520 is used to express the interaction between the input processing units, the number of processing units is not determined by a standard method, and it is often necessary to determine the best by experiment. The number, this hidden layer can be implemented using a nonlinear conversion function. Usually, the network can have more than one hidden layer or no hidden layer at all; its output layer 1530 is used to represent the output variable of the network, and the number of processing units depends on Depending on the problem, this output layer can be implemented using a nonlinear transfer function to represent the output. This conversion function can be used to control the formation of the output unit during the entire operation of the neural network. Therefore, according to the disclosed embodiment, for each physique _j , the evaluation index Y _j can be used as the inferential output value of the jth neuron of the output layer, that is, Y _j = β ₁ × pulse wave characteristic value + β _2j , j=1, 2, 3, ..., k. Accordingly, according to an embodiment of the present disclosure, the fifteenth figure is an example of the inverted transmission neural network architecture as an index risk assessment model. Figure 16 is a flow chart showing the operation of the inverse transfer type neural network of the fifteenth diagram, including steps 1610, 1620, 1630, ..., and 1680, in accordance with an embodiment of the present disclosure.

In step 1610, the transfer function and the network parameter values (learning rate η, inertia factor α) are set. In step 1620, the initial weighted value of the network and the initial bias value are set in a uniform random number. In step 1630, the training sample x _i and the target output value T _{j are} input. In step 1640, the first network are calculated as H _h h-th neuron of the hidden layer output, and the Y _j is the j-th output layer neuron output inference value. The output of the hidden layer is calculated as follows: The inferential output value of the jth neuron in the output layer is calculated as follows: Where x _i is the output value of the i-th neuron in the input layer, net _h is the weighted product sum of the h- th neuron in the hidden layer, net _j is the weighted product sum of the j- th neuron in the output layer, and f is the hidden layer and the transfer function of the output layer, W _hi is the input layer, the i-th neuron to the hidden layer h-th weighting value (weight) between neuronal function, W _hj hidden layer h-th neuron of the output layer of the j-th weighting value function between neurons, θ _h is the h-th partial weights neuron hidden layer, θ _j is the j-th partial weights neuron output layer, N _inp is the number of input neurons, N _hid hidden The number of layers of neurons.

In step 1650, the amount of difference between the output layer and the hidden layer is calculated. Calculated as follows: Wherein, δ _j is the j-gap output layer neurons, δ _h is the gap between the amount of the h-th hidden neuron layer, T _j is the j th value of the output target neuron.

In step 1660, the weighting value and the offset weight correction amount between the layers are calculated and calculated. The calculation formula is as follows: △ W _hj = ηδ _j H _h + α × △ W _hj , Δ θ _j = - ηδ _j + α × Δ θ _j , △ W _ih = ηδ _h X _i + α × △ W _ih , Δ θ _{h = - ηδ h + α ×} △ θh, where, △ W _hj hidden layer h-th neuron of the output layer of the j-th neural weighting value correction amount between the element, △ θ _j is the output layer, the j-th neuron The bias value correction amount, Δ W _ih is the weighted value correction amount between the i- th neuron in the input layer and the h- th neuron in the hidden layer, and Δ θ _h is the bias value correction amount of the h- th neuron in the hidden layer .

In step 1670, the weighting values and the bias values between the layers are updated. The calculation formula is as follows: W _hj = W _hj + Δ W _hj , θ _j = θ _j + Δ θ _j , W _ih = W _ih + Δ W _ih , θ _h = θ _h + Δ θ _h . In step 1680, steps 1630 through 1670 are repeated until the network converges. The purpose of network learning is to adjust the connection weighting value to minimize the error function of the network. Generally, the learning quality is adjusted by using the error function E (or energy function), which is defined as follows: Where N _cycle is the number of learning cycles, T _j is the target output value of the jth neuron, and Y _j is the inferential output value of the jth neuron of the output layer.

In summary, the embodiments of the present disclosure provide a pulse wave and physical health risk assessment system and method, which analyzes a user's pulse wave characteristic parameters and identifies a user's physique by extracting user pulse wave information. Category, according to different physical characteristics, define different pulse-risk models or risk criteria according to the body group. The user's disease risk profile and its corresponding disease risk criteria or risk model are then used to estimate the user's disease risk. It can provide users with accurate disease risk parameters, as well as awareness of self-health status, prevention of underestimating the risk of high-risk groups, and timely intervention to improve health.

The above is only the embodiment of the disclosure, and the scope of the disclosure is not limited thereto. All changes and modifications made to the scope of the present invention should remain within the scope of the present invention.

Disease risk curve of four ethnic groups R1~R4‧‧

200‧‧‧Health Risk Assessment System

210‧‧‧ Pulse wave sensing device

220‧‧‧Physical identification device

230‧‧‧ risk assessment device

212‧‧‧ Pulse wave characteristic parameters

222‧‧‧Physical classification

232‧‧‧ Disease Risk Information

H‧‧‧ pulse wave height

H1‧‧‧ main wave amplitude

h2‧‧‧Pre-pulse amplitude

h3‧‧‧Heavy wave amplitude

T1‧‧‧ acute ejection period

T2‧‧‧ Systolic

T‧‧‧ pulse period

f ' ( t ), f " ( t ), f ''' ( t ), f "" ( t ) ‧ ‧ one to four derivatives of pulse waves

C5 to C8‧‧‧ fifth to eighth modal functions

β ₁ , β ₂₁ , β ₂₂ , β ₂₂ ,. . . , β _2k ‧‧‧ parameters

P _i ‧‧‧The probability of getting the disease in the i-th sample

Logit P‧‧‧logistic function

k‧‧‧Number of categories of constitutional classification

1310‧‧‧ obtaining a set of pulse wave characteristic parameters of the at least one user by using a plurality of pulse signals of at least one user sensed by a pulse wave sensing device

1320‧‧‧Using a one-piece identification device to identify at least one integral state corresponding to the at least one user

1330‧‧‧ Using a risk assessment device to define respective disease risk criteria or disease risk models based on at least one user population corresponding to the at least one user

1340‧‧‧ estimating at least one disease risk of the at least one user by the at least one user of the set of pulse wave characteristics and the corresponding disease risk criterion or disease risk model by the risk assessment device degree

1400‧‧‧Health Risk Assessment

1410‧‧‧User test device

1420‧‧‧Physical classifier

1430‧‧‧ Pulse Analyzer

1440‧‧‧Physical database

1510‧‧‧Input layer

1520‧‧‧ hidden layer

1530‧‧‧ Output layer

x _i ‧‧‧ Output value of the i-th neuron in the input layer

Y _j ‧‧‧ is the inferential output value of the jth neuron in the output layer

1610‧‧‧Set conversion function and network parameter values

1620‧‧‧Set the initial weighting and initial bias of the network with a random random number value

1630‧‧‧Enter training sample x _i and target output value T _j

1640‧‧‧ Calculate H _h as the output of the hth neuron in the hidden layer, and Y _j as the inferential output value of the jth neuron in the output layer

1650‧‧‧ Calculate the difference between the output layer and the hidden layer

1660‧‧‧ Calculate and calculate the weighted value and the bias value correction between layers

1670‧‧‧Update weighted values and partial weights between layers

1680‧‧‧ Repeat steps 1630 through 1670 until the network converges

The first picture is a distribution of the pulse wave and disease risk of different ethnic groups.

The second figure illustrates a pulse wave and body health risk assessment system in accordance with an embodiment of the present disclosure.

The third figure is a pulse signal diagram illustrating a typical age group according to an embodiment of the present disclosure.

The fourth to fourth C diagrams are pulse wave signal diagrams of three age groups, respectively, according to an embodiment of the disclosure.

The fifth figure is an exploded view of a fourth derivative of a pulse wave signal according to an embodiment of the present disclosure.

The sixth figure is an exploded view of a pulse signal Hilbert-Huang conversion according to an embodiment of the present disclosure.

The seventh figure illustrates a constitution classification determination rule according to an embodiment of the present disclosure.

The eighth figure is an example of an index risk assessment model and an example of a logistic regression model as an index risk assessment model according to an embodiment of the present disclosure.

The ninth figure is a schematic diagram showing an example of defining different pulse wave-risk evaluation models under different constitutions according to an embodiment of the present disclosure.

The tenth figure is a schematic diagram showing an example of different risk levels according to the relationship between different physical qualities and pulse wave signals according to an embodiment of the present disclosure.

An eleventh diagram illustrates an evaluation rule in accordance with an embodiment of the present disclosure.

The twelfth figure is a schematic diagram showing an example of defining three different levels of risk, high, medium and low under different constitutions according to an embodiment of the present disclosure.

A thirteenth diagram is a flowchart of a pulse wave and physical health risk assessment method according to an embodiment of the present disclosure.

The fourteenth embodiment is an example of implementing a pulse wave and physique health risk assessment meter technique in accordance with an embodiment of the present disclosure.

The fifteenth figure is an example of using an inverted transfer type neural network architecture as an index risk assessment model according to an embodiment of the present disclosure.

Figure 16 is a flow chart showing the operation of the inverse transfer type neural network of the fifteenth diagram according to an embodiment of the present disclosure.

200‧‧‧Health Risk Assessment System

210‧‧‧ Pulse wave sensing device

220‧‧‧Physical identification device

230‧‧‧ risk assessment device

212‧‧‧ Pulse wave characteristic parameters

222‧‧‧Physical classification

232‧‧‧ Disease Risk Information

Claims (16)

A pulse wave and physical health risk assessment system includes: a pulse wave sensing device that senses a plurality of pulse signals of at least one user and defines one or more characteristics of the plurality of pulse wave signals to calculate the at least a set of pulse wave characteristic parameters corresponding to each user; the integrated quality identification device, identifying the physical fitness classification of the at least one user from the extracted body information of the at least one user; and a risk assessment The device estimates at least one disease risk information corresponding to at least one user according to the set of pulse wave characteristic parameters obtained by the plurality of pulse wave signals and the identified group of body categorizations, wherein the evaluation The set of pulse wave characteristic parameters and the group constitution classification are regarded as the input two sets of health risk parameters, and then an indicator risk assessment model is used to calculate the health risk assessment information related to the constitution. The system of claim 1, wherein the at least one disease risk information is one or more of the foregoing information of a disease risk index, a disease risk level, and a disease risk level. The system of claim 1, wherein the at least one of the at least one user identifies the at least one user's physique by using at least one questionnaire and at least one physiological signal acquisition method. classification. The system of claim 1, wherein the pulse wave sensing device parses the plurality of pulse wave signals via a pulse wave analyzer to calculate the set of pulse wave characteristic parameters. Such as the system described in claim 1, wherein the disease risk The newsletter contains at least one circulatory system disease risk information. The system of claim 2, wherein the risk assessment device gives the at least one user a different risk level value according to the relationship between the different physical qualities and the plurality of pulse signals. The system of claim 1, wherein the risk assessment device defines different pulse-risk assessment models according to different physical qualities in the case of the same pulse signal feature. The system of claim 4, wherein the pulse wave analyzer uses time domain or frequency domain pulse wave analysis to define the one or more features of the plurality of pulse signals. A pulse wave and physique health risk assessment method includes: sensing, by a pulse wave sensing device, a plurality of pulse signals of at least one user, and defining one or more characteristics of the plurality of pulse wave signals to calculate the Identifying, by the at least one user, a set of pulse wave characteristic parameters, and using the integrated quality identification device, identifying at least one user's at least one user's physical quality information from the extracted body; and Estimating at least one disease risk information corresponding to at least one user by using a risk assessment device; wherein the evaluating the set of pulse wave characteristic parameters and the set of physique classification corresponding to the at least one holistic state as the input two sets of health risks Parameters, and then use an indicator risk assessment model to calculate health-related health assessment information. The method of claim 9, wherein the method uses at least one questionnaire scale and at least one physiological signal acquisition method, At least one of the ways to identify the at least one quality state corresponding to at least one user. The method of claim 9, wherein the method defines a corresponding at least one user group according to the at least one entity state of the at least one user, and correspondingly according to the at least one user group definition Disease risk criteria or disease risk models. The method of claim 11, wherein the method defines different disease risk criteria or risk models for user groups of different constitutions corresponding to the same pulse wave parameters. The method of claim 11, wherein the disease risk criterion or the risk model is an empirical level and one of the foregoing ones. The method of claim 11, wherein the disease risk model is an inverse transfer type neural network and a logistic regression model, one of the foregoing. The method of claim 9, wherein the method parses the plurality of pulse signals using at least one numerical method to calculate the set of pulse wave characteristic parameters. The method of claim 9, wherein the at least one numerical method comprises time domain or frequency domain pulse wave analysis to define the one or more features of the plurality of pulse signals.