WO2021189982A1

WO2021189982A1 - Health information distribution determination method and apparatus based on machine learning

Info

Publication number: WO2021189982A1
Application number: PCT/CN2020/136368
Authority: WO
Inventors: 曾振; 王健宗; 程宁
Original assignee: 平安科技（深圳）有限公司
Priority date: 2020-10-26
Filing date: 2020-12-15
Publication date: 2021-09-30
Also published as: CN112364896A; CN112364896B

Abstract

A health information distribution determination method and apparatus based on machine learning, which relate to the technical field of data processing, and mainly aim to solve the problem of it being impossible to meet the convenience and rapidity requirements of health care for data processing due to the low determination efficiency of an existing health information distribution image. The method comprises: acquiring spectral data (101); on the basis of a trained spectral classification model, performing classification processing on the spectral data, so as to obtain a classification processing result that includes health features respectively marked with the spectral data (102), wherein the spectral classification model is a hybrid model established on the basis of a combination of different levels of machine learning models; and according to a preset spectral integration weight, performing integration processing on the health features in the classification processing result, so as to obtain a health information distribution image (103). The method is mainly used for determining a health information distribution on the basis of machine learning.

Description

Method and device for determining health information distribution based on machine learning

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on October 26, 2020, the application number is 202011153516.1, and the application title is "Method and Apparatus for Determining Health Information Distribution Based on Machine Learning", the entire content of which is incorporated by reference Incorporated in this application.

Technical field

This application relates to the field of data processing technology, in particular to a method and device for determining the distribution of health information based on machine learning.

Background technique

As people pay more and more attention to their own health and the health of others, intelligent health check-ups have gradually developed into the focus of medical insurance programs. Among them, the intelligent health checkup refers to obtaining the user's basic health data through simple medical examination methods such as blood, blood pressure, blood sugar, and ultrasound images, and analyzing the basic health data through accurate data processing methods to obtain the user's health indicators Or the distribution of various health information.

At present, the inventor realizes that the existing distribution of health information is usually based on the comparison of individual indicators in basic health data with international medical standards, which cannot meet the needs of comprehensive analysis of health information, and a single comparison method makes data processing obtainable. The results are more redundant; and basic monitoring data as a medical resource, a single comparison method cannot meet the need to determine the appropriate health information distribution image as different medical scenarios change, so that the health information distribution image determination efficiency Low, unable to meet the needs of health care for the convenience and speed of data processing.

technical problem

In view of this, the present application provides a method and device for determining the distribution of health information based on machine learning. The main purpose is to solve the problem that the existing health information distribution image determination efficiency is low, which cannot meet the convenience and rapidity of data processing in health care. The question of demand.

Technical solutions

According to one aspect of this application, a method for determining the distribution of health information based on machine learning is provided, which includes:

Obtain spectral data;

Performing classification processing on the spectrum data based on the trained spectrum classification model to obtain classification processing results containing the spectrum data respectively labeled health characteristics, the spectrum classification model being a hybrid model established based on a combination of machine learning models of different levels;

According to the preset spectral integration weights, the health features in the classification processing result are integrated to obtain a distribution image of health information.

According to another aspect of the present application, a device for determining the distribution of health information based on machine learning is provided, including:

Obtaining module for obtaining spectral data;

The classification processing module is used to classify the spectrum data based on the trained spectrum classification model to obtain a classification processing result including the health characteristics of the spectrum data respectively labeled, and the spectrum classification model is a combination of machine learning models based on different levels Established hybrid model;

The integration processing module is used to perform integration processing on the health features in the classification processing result according to preset spectral integration weights to obtain a distribution image of health information.

According to another aspect of the present application, a storage medium is provided, the storage medium stores at least one executable instruction, and the executable instruction causes a processor to execute a method for determining the distribution of health information based on machine learning, Wherein, the method for determining the distribution of health information based on machine learning includes the following steps:

Obtain spectral data;

According to another aspect of the present application, a computer device is provided, including: a processor, a memory, a communication interface, and a communication bus. The processor, the memory, and the communication interface complete mutual communication through the communication bus. Communication

The memory is used to store at least one executable instruction that causes the processor to execute a method for determining the distribution of health information based on machine learning, wherein the method for determining the distribution of health information based on machine learning is It includes the following steps:

Obtain spectral data;

Beneficial effect

This application provides a method and device for determining the distribution of health information based on machine learning. Compared with the prior art, it satisfies the demand for determining the distribution of health information in a health checkup, and determines the health characteristics of users more efficiently and accurately. This greatly meets the needs of the health and medical field for the convenience and speed of data processing.

Description of the drawings

By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only used for the purpose of illustrating the preferred embodiments, and are not considered as a limitation to the application. Also, throughout the drawings, the same reference symbols are used to denote the same components. In the attached picture:

FIG. 1 shows a flowchart of a method for determining the distribution of health information based on machine learning provided by an embodiment of the present application;

FIG. 2 shows a flowchart of another method for determining the distribution of health information based on machine learning provided by an embodiment of the present application;

Figure 3 shows a block diagram of a device for determining the distribution of health information based on machine learning provided by an embodiment of the present application;

FIG. 4 shows a block diagram of another device for determining the distribution of health information based on machine learning provided by an embodiment of the present application;

Fig. 5 shows a schematic structural diagram of a computer device provided by an embodiment of the present application.

The best mode of the present invention

Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the drawings show exemplary embodiments of the present disclosure, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

The embodiment of the application provides a method for determining the distribution of health information based on machine learning. As shown in FIG. 1, the method includes:

101. Acquire spectral data.

Among them, during the health checkup, by sampling the user's blood and saliva, and using a spectrometer to perform spectral detection on blood samples and saliva samples, the results include at least blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data. The spectrum data is used to carry out the distribution image of health information.

102. Perform classification processing on the spectrum data based on the trained spectrum classification model, and obtain a classification processing result including the health characteristics of the spectrum data respectively labeled.

Wherein, the spectral classification model is a hybrid model established based on a combination of machine learning models at different levels. In order to classify spectral data, the hybrid model in the embodiment of the present application is a machine learning model that includes two different classification functions, and is based on different types of machine learning models. Combine hierarchies. Specifically, combining different levels to establish a hybrid model can establish a hybrid relationship with the second classification model by using the first classification model as the input layer of the second classification model, where the first classification model and the second classification model The model is a different machine learning model. For example, the first classification model is a decision tree model, and the second classification model can be another classification model other than the decision data model, which serves as the output layer in the mixed model. This embodiment does not Make specific restrictions.

103. Perform integration processing on the health features in the classification processing result according to the preset spectral integration weights to obtain a distribution image of health information.

For the embodiments of the present application, in order to intelligently complete the determination of the distribution image of the health information, after the spectral classification model is used for classification processing, the classification processing result includes the health features of the spectral data mark probability, and the health features are used to represent different health The characteristic data on which the state is based, for example, the uric acid value may be characteristic data that characterizes rheumatic diseases, and human chorionic gonadotropin may be characteristic data that characterizes the state of pregnancy, etc., which are not specifically limited in the embodiments of the present application. Therefore, in order to make the classified health characteristics suitable for health examinations, so as to quickly, conveniently and accurately determine the user's health information, the spectral data of the classified different health characteristics are integrated and processed to obtain the distribution image of the health information , So as to meet the needs of health care for the convenience and speed of data processing.

It should be noted that the preset spectral integration weights are pre-configured for different health characteristics with respect to the distribution of spectral characteristics. For example, the weight of the health characteristics of myocarditis classified based on infrared spectrum data is 0.2, and the classification of ultraviolet spectrum data is The weight of the health feature of myocarditis is 0.4, so that the health information of the health feature of myocarditis is integrated based on the weights of 0.2 and 0.4 to obtain the distribution image of myocarditis, which is not specifically limited in the embodiment of the present application.

This application provides a method for determining the distribution of health information based on machine learning. Compared with the prior art, the embodiment of this application obtains spectral data; classifies the spectral data based on a trained spectral classification model to obtain Contains the classification processing results of the spectral data respectively labeling the health characteristics, the spectral classification model is a hybrid model established based on a combination of machine learning models of different levels; according to the preset spectral integration weights, the health characteristics in the classification processing results Perform integrated processing to obtain distribution images of health information, meet the needs for determining the distribution of health information in health examinations, and determine the health characteristics of users more efficiently and accurately, thereby greatly satisfying the convenience and convenience of data processing in the health care field. The need for rapidity.

The embodiment of the present application provides another method for determining the distribution of health information based on machine learning. As shown in FIG. 2, the method includes:

201. Obtain a spectral training data set.

In the embodiments of the present application, in order to train the hybrid model and obtain the ability to accurately classify the spectral data, a spectral training data set is acquired, so as to obtain training data from the spectral training data set to train the hybrid model. Wherein, the spectral training data set includes spectral data corresponding to health features of different classifications, and the health features are feature data used to characterize different health states. In addition, the spectrum data includes at least blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, saliva ultraviolet spectrum data, and the spectrum data is characterized by light wavelength and amplitude. For example, for blood infrared spectrum data, the wavelength is λ The radiation quantity in the range of the small wavelength width dλ as the center is dX, then the radiation quantity corresponding to the unit wavelength interval is called the spectral density Xλ, that is, Xλ=dX/dλ, where the radiation quantity can be radiant flux, radiation Intensity, radiance, irradiance, etc. Generally speaking. Different wavelengths have different corresponding spectral densities. When the corresponding relationship between the spectral density of the light source and the wavelength is expressed as a function, the function is called the spectral distribution of the light source Xλ(λ), which is blood The infrared spectrum data is not specifically limited in the embodiment of this application.

202. Combine and construct a spectral classification model including at least two decision tree models and one neural network model.

For the embodiments of the present application, in order to improve the classification and processing capability of spectral data, thereby efficiently performing data classification processing, a spectral classification model of at least two decision tree models and one neural network model is constructed by combining. Wherein, the combination construction is realized by using the at least two decision tree models as the input layer and the one neural network model as the output layer. In the embodiment of the present application, since the spectral data may at least include blood infrared spectral data , Blood ultraviolet spectrum data, saliva infrared spectrum data, saliva ultraviolet spectrum data, in order to optimize the accuracy of the classification processing, at least two decision tree models are used as input levels, and a neural network model is used as the output layer to construct a mixed spectrum classification model.

It should be noted that different spectral data may lead to the determination of different health characteristics. Therefore, the first level is to establish at least two decision tree models. The specific steps include: using the Bootstrapping method from the original training set to randomly replace the sampling to select m Samples, a total of n_tree times are sampled, and n_tree training sets are generated. For n_tree training sets, we train n_tree decision tree models respectively. For a single decision tree model, the number of training sample features is, no pruning is required during the splitting process of the decision tree. The steps are:

A. Feature set D'={z1, z2, z3, z4}, the health features are classified into 2 categories, and the classification result is yes or no. For example, a decision tree for cold features can be constructed as the first layer of feature judgment as a spectrum Whether the data conforms to the cold characteristic spectral distribution between cb, if so, the second-level feature judges whether the a spectral data conforms to the viral cold spectral distribution between ft, and so on. Assuming that the given training set is D={(x1, y1), (x2, y2),,, (xNyN)}, the jth traversal xj and its value s can be selected as the segmentation traversal and segmentation point, definition 2 regions, R ₁ (j,s)={x|x _j ≤s}, R ₂ (j,s)={x|x _j ＞s}, then find the optimal segmentation traverse xj and the optimal cut Point s, solve

Among them, cm is the output of the decision tree on Rm, and is the mean value of the output yi corresponding to all input instances xi on the region Rm.

Repeat the above process for each region R1 and R2, until the stop condition is met, divide the input space into M regions R1, R2,,, RM to generate a decision tree:

B. For the second-class classification problem, if the probability that the sample point belongs to the first category is p, then the Gini index of the probability distribution is:

For sample set D, its Gini index is:

Among them, Ck is the sample set belonging to the k-th category in D, and K is the number of categories. If the sample set D is divided into two parts D1 and D2 according to whether the feature D'takes a certain possible value z, that is, D ₁ = { (x,y)|D'(x)=z}, D ₂ =DD ₁ , under the condition of characteristic D', the Gini index of set D is:

C. Decision tree generation steps: 1. Set the training data set of the node as D, and for each feature D'={z1, z2, z3, z4}, for each possible value {z1, z2, z3 , Z4}, according to the sample point {z1, z2, z3, z4} test is "yes" or "no" to divide D into two parts D1 and D2, and calculate

2. Among all possible features D'and all possible segmentation points {z1, z2, z3, z4}, select the feature with the smallest Gini index and its corresponding segmentation point as the optimal feature and optimal segmentation point. According to this, two sub-nodes are generated from the current node, and the training data set is allocated to the two sub-nodes according to the characteristics. 3. Recursively call 1. and 2. to the two sub-nodes until the stop condition is met to generate the CART decision tree n_tree. The above process is the generation process of a single decision tree model, and the generation process of at least two decision tree models They are all the same, so I won’t repeat them here.

In addition, after completing the establishment of multiple decision tree models, obtain the label probability matching the classification results of the health characteristics of different spectral data in the training set, and use the vector form as the training sample data of the neural network to train the neural network, for example, as a decision tree The classification result of is that the spectral data a is a feature of virus type 1 cold, and the spectral data a is a feature of rheumatism and a feature of meningeal inflammation. Construct a vector of full disease features, such as 150 disease vectors, corresponding to the above three health features configured as 1, and the rest are 0. When constructing neural network training samples, the input sample data is the label probability, that is, the classification of the health features is performed To determine the risk distribution, the label probability includes the weight configuration of at least 30 major disease features, the weight configuration of 80 medium disease features, and the weight configuration of 40 mild disease features. Therefore, the neural network training is performed to obtain The determination results of health characteristics including low-risk, medium-risk, and high-risk are not specifically limited in the embodiment of the present application.

203. Train the spectral classification model constructed by the combination based on the spectral training data set.

In the embodiments of the present application, in order to achieve classification training on spectral data, the combined-built spectral classification model is trained based on the training data in the spectral training data set to obtain a spectral classification model suitable for health feature classification after the training is completed.

204. Acquire spectral data.

Further, for further definition and explanation, step 204 may specifically include: acquiring spectrum data including at least blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data.

In the embodiments of the present application, in order to simplify the steps of determining the distribution of health information and improve the convenience of the user's health sample collection operation, when blood samples and saliva samples are collected, a spectrometer is used to perform spectral analysis on the blood samples and saliva samples. Output spectrum data including at least blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data. Of course, with the development of spectrometers, the analyzed spectrum data can also include non-ultraviolet and infrared spectrum data , The embodiments of this application do not make specific limitations.

The spectral data analyzed by the spectrometer is determined based on the wavelength and amplitude of different light rays in blood and saliva, thereby obtaining spectral data of different light rays such as ultraviolet and infrared rays, which are not specifically limited in the embodiments of the present application.

205. Perform classification processing on the spectrum data based on the trained spectrum classification model to obtain a classification processing result that includes the health characteristics of the spectrum data respectively labeled.

Further, in order to optimize the spectral data data and avoid abnormal data in the parsed spectral data from affecting the classification process, it is necessary to perform data preprocessing on the spectral data. In the embodiment of the present application, the spectral classification model is based on a trained spectrum. Before performing classification processing on the spectrum data, the method further includes: separately determining the wavelength values in the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data. Whether the amplitude value has a distorted state; if there is a distorted state, the wavelength value and amplitude value in the distorted state are filtered, and the filtered blood infrared spectrum data, the blood ultraviolet spectrum data, and the saliva infrared The spectrum data and the saliva ultraviolet spectrum data are used as the spectrum data to be classified.

For the embodiments of this application, since blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data are all embodied based on wavelength values and amplitude values, in order to filter abnormal data, determine blood Whether the wavelength value and amplitude value in the infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data are distorted. Among them, the distortion state is a sharp increase or decrease in the wavelength value and amplitude value. Generally, the surge or sharp decrease is configured to a distortion range that matches the normal wavelength value and amplitude value of the spectrum. If the distortion range exceeds this distortion range, it is determined that there is a distortion state. Perform filtering processing on the distorted wavelength value and amplitude value. The filtering process is to delete the distorted wavelength value, the blood infrared spectrum data corresponding to the amplitude value, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data. . Among them, if the wavelength value and amplitude value of any one of the spectrum data is distorted, the corresponding spectrum data will be deleted. If all the spectrum data is distorted, it means that the spectrometer has collected errors. You can delete all of them to re-analyze the blood and saliva samples. , The embodiments of this application do not make specific limitations.

206. Perform integration processing on the health features in the classification processing result according to the preset spectral integration weights to obtain a distribution image of health information.

For the embodiment of the present application, for further explanation and refinement, step 206 may specifically include: using a weighted sum method combined with preset spectral integration weights to count the blood infrared spectrum data and the blood ultraviolet spectrum in the classification processing result. Data, the saliva infrared spectrum data, and the integration interval of the health characteristics marked by the saliva ultraviolet spectrum data; the distribution image containing the health information of the integration interval is drawn in a superimposed manner.

In the embodiment of this application, since the classification processing results include low-risk, medium-risk, and high-risk health characteristics, in order to accurately obtain the distribution of health information, the weighted summation method is combined with preset spectral integration weight statistics to undergo classification processing. The integrated interval of the health characteristics of different levels of risk, and the distribution image of the health information of the integrated interval is drawn. Wherein, the preset spectral integration weights are pre-configured weights for different health characteristics with respect to the distribution of spectral characteristics. For example, blood ultraviolet spectrum data is classified into medium-risk virus type 1 cold characteristics, low-risk meningitis characteristics, and blood infrared Spectral data is classified into high-risk virus type 1 cold features and medium-risk meningitis features. Correspondingly, the blood ultraviolet spectrum data classifies the virus type 1 cold feature as 0.1, the meningitis feature as 0.6, and the blood infrared spectrum data classifies the virus type 1 If the cold feature is 0.3 and the meningitis feature is 0.3, then each weight is used for weighted summation, for example, 0.1*medium risk+0.3*high risk, where each level of risk is digitized in advance, and the corresponding weighted summation is configured After the numerical area of each risk, the weighted and summed health information of virus type 1 cold features is obtained, such as 0.1*medium risk+0.3*high risk→high risk, which is the high risk of virus 1 cold feature. Examples of this application There are no specific restrictions on the numerical value and numerical value area.

It should be noted that, in the distribution image of the health information containing the integration interval drawn in a superimposed manner, the integration interval is the risk interval in which different health characteristics are located. For example, the health information determined by the user may include a health characteristic or It includes multiple health features. Therefore, in order to visualize it in a unified manner, the distribution image is drawn in a superimposed manner. Each risk area in the distribution image can show the distribution of different health characteristics in an overlapping manner. For example, the medium risk area may include rheumatism. Features, meningitis features, in order to easily complete the distribution of health information.

Further, in order to meet the visualization requirements of the distribution of health information, after step 206, the embodiment of the present application further includes: after receiving a query request for a distribution image of health information, extracting historical images matching the distribution image; according to different colors Rendering the distributed image and the historical image, and rendering the distributed image and the historical image in a semi-transparent overlapping manner for output.

For the embodiment of this application, in order to meet the management needs of health information, when combined with the distribution image request of the health information, the historical image matching the distribution image is extracted, that is, the distribution image generated by the user's historical health characteristics, according to different colors Render the distributed image and the historical image, and combine the rendered distributed image and the historical image in a semi-transparent overlapping manner to output, so that the user can view the historical image and the distributed image of different colors through the semi-transparent rendering image. Among them, if there are multiple historical images, multiple colors can be rendered after labeling according to time information, thereby improving the visualization effect.

This application provides another method for determining the distribution of health information based on machine learning. Compared with the prior art, the embodiment of this application obtains spectral data; and classifies the spectral data based on a trained spectral classification model. Obtain a classification processing result including the spectrum data separately labeled health characteristics, the spectrum classification model is a hybrid model established based on a combination of machine learning models of different levels; according to preset spectrum integration weights, the health in the classification processing results The characteristics are integrated and processed to obtain the distribution image of health information, which meets the demand for determining the distribution of health information in the health examination, and more efficiently and accurately determines the health characteristics of the user, thereby greatly satisfying the convenience of data processing in the health care field , The need for rapidity.

Further, as an implementation of the method shown in FIG. 1, an embodiment of the present application provides a device for determining the distribution of health information based on machine learning. As shown in FIG. 3, the device includes:

The obtaining module 31 is used to obtain spectral data;

The classification processing module 32 is configured to classify the spectrum data based on the trained spectrum classification model to obtain a classification processing result including the health characteristics of the spectrum data respectively labeled, and the spectrum classification model is a machine learning model based on different levels Hybrid model established by combination;

The integration processing module 33 is configured to perform integration processing on the health features in the classification processing result according to preset spectral integration weights to obtain a distribution image of health information.

This application provides a device for determining the distribution of health information based on machine learning. Compared with the prior art, the embodiment of this application obtains spectral data; classifies the spectral data based on a trained spectral classification model to obtain Contains the classification processing results of the spectral data respectively labeling the health characteristics, the spectral classification model is a hybrid model established based on a combination of machine learning models of different levels; according to the preset spectral integration weights, the health characteristics in the classification processing results Perform integrated processing to obtain distribution images of health information, meet the needs for determining the distribution of health information in health examinations, and determine the health characteristics of users more efficiently and accurately, thereby greatly satisfying the convenience and convenience of data processing in the health care field. The need for rapidity.

Further, as an implementation of the method shown in FIG. 2, an embodiment of the present application provides another device for determining the distribution of health information based on machine learning. As shown in FIG. 4, the device includes:

The obtaining module 41 is used to obtain spectral data;

The classification processing module 42 is configured to classify the spectrum data based on the trained spectrum classification model to obtain a classification processing result including the health characteristics of the spectrum data respectively labeled, and the spectrum classification model is a machine learning model based on different levels Hybrid model established by combination;

The integration processing module 43 is configured to perform integration processing on the health features in the classification processing result according to preset spectral integration weights to obtain a distribution image of health information.

Further, the device further includes: a construction model 44, a training model 45,

The acquiring module 41 is further configured to acquire a spectral training data set, the spectral training data set including the spectral data corresponding to the health characteristics of different classifications;

The construction module 44 is configured to construct a spectral classification model including at least two decision tree models and one neural network model in combination, wherein the combination is constructed by using the at least two decision tree models as input layers and Describe a neural network model implemented for the output layer;

The training module 45 is configured to train the combined spectral classification model based on the spectral training data set.

Further, the acquisition module 41 is specifically configured to acquire spectrum data including at least blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data.

Further, the integration processing module 43 includes:

The statistics unit 4301 is configured to use a weighted sum method combined with preset spectral integration weights to count the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data in the classification processing result. The integration interval of the health characteristics marked by the spectral data;

The drawing unit 4302 is configured to draw a distribution image containing the health information of the integration interval in a superimposed manner.

Further, the device further includes:

The judging module 46 is configured to respectively judge whether the wavelength value and the amplitude value in the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data are in a distorted state;

The filter processing module 47 is configured to filter the wavelength value and amplitude value in the distortion state if there is a distortion state, and combine the filtered blood infrared spectrum data, the blood ultraviolet spectrum data, and the saliva infrared spectrum data. Spectral data, the saliva ultraviolet spectrum data are used as spectrum data to be classified, and the filtering process is to delete the distorted wavelength value, the blood infrared spectrum data corresponding to the amplitude value, blood ultraviolet spectrum data, and saliva infrared spectrum Data, saliva UV spectrum data.

Further, the integration processing module 43 includes:

The extracting unit 4303 is configured to extract a historical image matching the distributed image after receiving a query request for a distributed image of health information;

The output unit 4304 is configured to render the distributed image and the historical image according to different colors, and combine and render the distributed image and the historical image in a semi-transparent overlapping manner for output.

Further, the health characteristics are characteristic data used to characterize different health states.

This application provides another device for determining the distribution of health information based on machine learning. Compared with the prior art, the embodiment of this application obtains spectral data; and classifies the spectral data based on a trained spectral classification model. Obtain a classification processing result including the spectrum data separately labeled health characteristics, the spectrum classification model is a hybrid model established based on a combination of machine learning models of different levels; according to preset spectrum integration weights, the health in the classification processing results The characteristics are integrated and processed to obtain the distribution image of health information, which meets the demand for determining the distribution of health information in the health examination, and more efficiently and accurately determines the health characteristics of the user, thereby greatly satisfying the convenience of data processing in the health care field , The need for rapidity.

According to an embodiment of the present application, a storage medium is provided. The storage medium may be non-volatile or volatile. The storage medium stores at least one executable instruction, and the computer executable instruction can execute the foregoing In the method for determining the distribution of health information based on machine learning in any method embodiment, the method for determining the distribution of health information based on machine learning includes the following steps:

Obtain spectral data;

FIG. 5 shows a schematic structural diagram of a computer device according to an embodiment of the present application, and the specific embodiment of the present application does not limit the specific implementation of the computer device.

As shown in FIG. 5, the computer device may include: a processor (processor) 502, a communication interface (Communications Interface) 504, a memory (memory) 506, and a communication bus 508.

Among them, the processor 502, the communication interface 504, and the memory 506 communicate with each other through the communication bus 508.

The communication interface 504 is used to communicate with other devices, such as network elements such as clients or other servers.

The processor 502 is configured to execute the program 510, and specifically can execute the relevant steps in the foregoing embodiment of the method for determining the distribution of health information based on machine learning.

Specifically, the program 510 may include program code, and the program code includes a computer operation instruction.

The processor 502 may be a central processing unit CPU, or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the computer device may be the same type of processor, such as one or more CPUs, or different types of processors, such as one or more CPUs and one or more ASICs.

The memory 506 is used to store the program 510. The memory 506 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), for example, at least one disk memory.

The program 510 may be specifically used to cause the processor 502 to perform the following operations:

Obtain spectral data;

Obviously, those skilled in the art should understand that the above-mentioned modules or steps of this application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed images are composed of multiple computing devices. On the network, they can optionally be implemented with program codes executable by a computing device, so that they can be stored in a storage device for execution by the computing device, and in some cases, they can be different from those here. The steps shown or described are executed in sequence, or they are respectively fabricated into individual integrated circuit modules, or multiple modules or steps of them are fabricated into a single integrated circuit module for implementation. In this way, this application is not limited to any specific combination of hardware and software.

The above descriptions are only preferred embodiments of the application, and are not intended to limit the application. For those skilled in the art, the application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application shall be included in the protection scope of this application.

Claims

A method for determining the distribution of health information based on machine learning, which includes:

Obtain spectral data;

Performing classification processing on the spectrum data based on the trained spectrum classification model to obtain classification processing results containing the spectrum data respectively labeled health characteristics, the spectrum classification model being a hybrid model established based on a combination of machine learning models of different levels;

According to the preset spectral integration weights, the health features in the classification processing result are integrated to obtain a distribution image of health information.
The method according to claim 1, wherein, before the spectral data is classified based on the trained spectral classification model, the method further comprises:

Acquiring a spectral training data set, the spectral training data set including the spectral data corresponding to the health features of different classifications;

The combined construction includes a spectral classification model including at least two decision tree models and one neural network model, wherein the combined construction is constructed with the at least two decision tree models as the input layer and the one neural network model as the output layer Implemented

Based on the spectral training data set, the spectral classification model constructed by the combination is trained.
The method according to claim 1 or 2, wherein said acquiring spectral data comprises:

Obtain spectrum data including at least blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data.
The method according to claim 3, wherein said integrating the health features in the classification processing result according to a preset spectral integration weight to obtain a distribution image of health information comprises:

Using a weighted sum method combined with preset spectral integration weights to count the health characteristics of the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data in the classification processing result ’S integration interval;

The distribution image containing the health information of the integration interval is drawn in a superimposed manner.
The method according to claim 4, wherein, before the spectral data is classified based on the trained spectral classification model, the method further comprises:

Separately determining whether the wavelength value and amplitude value in the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data are in a distorted state;

If there is a distortion state, the wavelength value and amplitude value in the distortion state are filtered, and the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data are filtered. The spectrum data is used as the spectrum data to be classified, and the filtering process is to delete the blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data corresponding to the distorted wavelength value and amplitude value.
4. The method according to claim 4, wherein after the rendering of the distribution image containing the health information of the integration interval in a superimposed manner, the method further comprises:

After receiving the distributed image query request for health information, extract historical images matching the distributed image;

The distributed image and the historical image are rendered according to different colors, and the distributed image and the historical image are combined and rendered in a semi-transparent overlapping manner for output.
The method according to any one of claims 1 to 6, wherein the health characteristics are characteristic data used to characterize different health states.
A device for determining the distribution of health information based on machine learning, which includes:

Obtaining module for obtaining spectral data;

The classification processing module is used to classify the spectrum data based on the trained spectrum classification model to obtain a classification processing result including the health characteristics of the spectrum data respectively labeled, and the spectrum classification model is a combination of machine learning models based on different levels Established hybrid model;

The integration processing module is used to perform integration processing on the health features in the classification processing result according to preset spectral integration weights to obtain a distribution image of health information.
A storage medium storing at least one executable instruction, the executable instruction causing a processor to execute a method for determining the distribution of health information based on machine learning:

Wherein, the method for determining the distribution of health information based on machine learning includes:

Obtain spectral data;

Performing classification processing on the spectrum data based on the trained spectrum classification model to obtain classification processing results containing the spectrum data respectively labeled health characteristics, the spectrum classification model being a hybrid model established based on a combination of machine learning models of different levels;

According to the preset spectral integration weights, the health features in the classification processing result are integrated to obtain a distribution image of health information.
The storage medium according to claim 9, wherein before the classification processing of the spectral data based on the trained spectral classification model, the method further comprises:

Acquiring a spectral training data set, the spectral training data set including the spectral data corresponding to the health features of different classifications;

The combined construction includes a spectral classification model including at least two decision tree models and one neural network model, wherein the combined construction is constructed with the at least two decision tree models as the input layer and the one neural network model as the output layer Implemented

Based on the spectral training data set, the spectral classification model constructed by the combination is trained.
The storage medium according to claim 9 or 10, wherein said acquiring spectral data comprises:

Obtain spectrum data including at least blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data.
11. The storage medium according to claim 11, wherein said performing integration processing on the health features in the classification processing result according to preset spectral integration weights to obtain a distribution image of health information comprises:

Using a weighted sum method combined with preset spectral integration weights to count the health characteristics of the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data in the classification processing result ’S integration interval;

The distribution image containing the health information of the integration interval is drawn in a superimposed manner.
The storage medium according to claim 12, wherein, before the spectral data is classified based on the trained spectral classification model, the method further comprises:

Separately determining whether the wavelength value and amplitude value in the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data are in a distorted state;

If there is a distortion state, the wavelength value and amplitude value in the distortion state are filtered, and the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data are filtered. The spectrum data is used as the spectrum data to be classified, and the filtering process is to delete the blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data corresponding to the distorted wavelength value and amplitude value.
11. The storage medium according to claim 12, wherein after said drawing the distribution image containing the health information of the integration interval in a superimposed manner, the method further comprises:

After receiving the distributed image query request for health information, extract historical images matching the distributed image;

The distributed image and the historical image are rendered according to different colors, and the distributed image and the historical image are combined and rendered in a semi-transparent overlapping manner for output.
14. The storage medium according to any one of claims 9-14, wherein the health characteristics are characteristic data used to characterize different health states.
A computer device includes: a processor, a memory, a communication interface, and a communication bus. The processor, the memory, and the communication interface communicate with each other through the communication bus;

The memory is used to store at least one executable instruction that causes the processor to execute a method for determining the distribution of health information based on machine learning, wherein the method for determining the distribution of health information based on machine learning is It includes the following steps:

Obtain spectral data;

Performing classification processing on the spectrum data based on the trained spectrum classification model to obtain classification processing results containing the spectrum data respectively labeled health characteristics, the spectrum classification model being a hybrid model established based on a combination of machine learning models of different levels;

According to the preset spectral integration weights, the health features in the classification processing result are integrated to obtain a distribution image of health information.
The computer device according to claim 16, wherein before the classification processing of the spectrum data based on the trained spectrum classification model, the method further comprises:

Acquiring a spectral training data set, the spectral training data set including the spectral data corresponding to the health features of different classifications;

The combined construction includes a spectral classification model including at least two decision tree models and one neural network model, wherein the combined construction is constructed with the at least two decision tree models as the input layer and the one neural network model as the output layer Implemented

Based on the spectral training data set, the spectral classification model constructed by the combination is trained.
The computer device according to claim 16 or 17, wherein said acquiring spectral data comprises:

Obtain spectrum data including at least blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data.
18. The computer device according to claim 18, wherein said performing integration processing on the health features in the classification processing result according to preset spectral integration weights to obtain a distribution image of health information comprises:

Using a weighted sum method combined with preset spectral integration weights to count the health characteristics of the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data in the classification processing result ’S integration interval;

The distribution image containing the health information of the integration interval is drawn in a superimposed manner.
The computer device according to claim 19, wherein, before the classification processing of the spectrum data based on the trained spectrum classification model, the method further comprises:

Separately determining whether the wavelength value and amplitude value in the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data are in a distorted state;

If there is a distortion state, the wavelength value and amplitude value in the distortion state are filtered, and the blood infrared spectrum data, the blood ultraviolet spectrum data, the saliva infrared spectrum data, and the saliva ultraviolet spectrum data are filtered. The spectrum data is used as the spectrum data to be classified, and the filtering process is to delete the blood infrared spectrum data, blood ultraviolet spectrum data, saliva infrared spectrum data, and saliva ultraviolet spectrum data corresponding to the distorted wavelength value and amplitude value.