CN113851221A - Dynamic health evaluation method and system based on time sequence body measurement data - Google Patents
Dynamic health evaluation method and system based on time sequence body measurement data Download PDFInfo
- Publication number
- CN113851221A CN113851221A CN202110945416.0A CN202110945416A CN113851221A CN 113851221 A CN113851221 A CN 113851221A CN 202110945416 A CN202110945416 A CN 202110945416A CN 113851221 A CN113851221 A CN 113851221A
- Authority
- CN
- China
- Prior art keywords
- evaluation index
- regression equation
- measurement data
- index attribute
- moment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Epidemiology (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention provides a dynamic health evaluation method and system based on time sequence body measurement data, and relates to the technical field of medical health evaluation. Firstly, determining an evaluation index attribute set of an original physical measurement data dataset, and carrying out differentiation classification on the original physical measurement data dataset based on the evaluation index attribute set; then, a stepwise regression equation is constructed for the evaluation index attribute subset and the data subset at each moment; and finally merging the stepwise regression equations of the data set at all the moments by using a weighted moving average method based on the time sequence characteristics of the body measurement data set to obtain a final weighted regression equation. And then the final weighted regression equation can be used for carrying out dynamic health evaluation on different patients based on time series physical measurement data. The method solves the problem that the health evaluation result is influenced due to single information source and strong subjectivity of the body measurement data, and simultaneously solves the problem that the existing dynamic health evaluation technology is poor in generalization.
Description
Technical Field
The invention relates to the technical field of medical health evaluation, in particular to a dynamic health evaluation method and system based on time sequence body measurement data.
Background
In the medical field, many medical physical measurement data exist in a time series form, such as continuous monitoring data of blood sugar, blood pressure, blood fat and the like related to chronic diseases, data of heartbeat, pulse and body position change continuously monitored by intelligent wearable equipment and the like. Aiming at health time sequence data with strong scale and structural characteristics, potential modes and valuable information can be explored through different analysis methods, then valuable prediction is made on the future trend of the health time sequence data according to the mined potential rules, valuable data support can be provided for health evaluation and prejudgment of patients, and the method is significant.
At present, health evaluation is mostly realized by methods such as feature scoring, health degree evaluation or self-testing health evaluation sample tables (SRHMS) and by adopting a time series model according to the characteristics of time series data and related health information contained in health time series data; corresponding health assessment is given by adopting a health assessment method for listing characteristics or selecting scores by using a health assessment scale, and corresponding health assessment is obtained by adopting an SVM algorithm and a BP neural network or utilizing algorithms such as big data analysis and the like to predict the illness state of a patient.
However, although these methods can provide scientific basis and data support for doctor diagnosis to some extent, the following problems still exist: on one hand, the subjective survey method and the physiological parameter detection method adopted by the existing human health evaluation technology have subjective and single information sources, so that the health evaluation result is inaccurate; on the other hand, the original technical scheme only analyzes a certain current medical scene or disease, different characteristics are difficult to screen and multiplex, and the health evaluation model has no usability and generalization.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides a dynamic health evaluation method and system based on time sequence body measurement data, and solves the problems of inaccurate evaluation result and poor generalization of the existing health evaluation method.
(II) technical scheme
In order to achieve the purpose, the invention is realized by the following technical scheme:
in a first aspect, the present invention first provides a dynamic health evaluation method based on time series physical measurement data, where the method includes:
acquiring an evaluation index attribute set based on the acquired original physical measurement data set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation among evaluation index attributes; differentiating and classifying the raw physical measurement dataset into a plurality of data subsets based on the evaluation index attribute subsets;
constructing a stepwise regression equation based on the evaluation index attribute subset and the data subset corresponding to each moment of the original physical measurement data set;
merging stepwise regression equations at all moments by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set to obtain a final weighted regression equation;
and performing dynamic health evaluation by using a final weighted regression equation based on the time sequence body measurement data to be evaluated.
Preferably, the determining an evaluation index attribute set of the time series physical measurement data, and acquiring a partitioned data set based on the evaluation index attribute set includes:
s11, determining the evaluation index attribute of the original body measurement data required by the construction of the stepwise regression equation based on the distribution characteristics and the correlation of the original body measurement data set, forming an evaluation index attribute set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets according to the strong correlation among the evaluation index attributes;
s12, calculating the quantitative result of the evaluation index attribute in the evaluation index attribute subset;
and S13, classifying the original body measurement data set based on the quantification result to obtain a plurality of data subsets.
Preferably, the evaluation index attributes include height and weight;
the quantitative result of the evaluation index attribute includes a BMI index:
wherein the content of the first and second substances,
represents body weight;h represents height.
Preferably, the constructing a stepwise regression equation based on the evaluation index attribute subset and the data subset corresponding to each time of the original physical measurement data set includes:
and constructing a stepwise regression equation by taking the evaluation index attribute subset at each moment as a potential item, wherein the process of constructing the stepwise regression equation specifically comprises the following steps:
s21, constructing a stepwise regression initial equation at the current moment;
s22, if a certain potential item which is not added into the model exists and the p value of the F statistic is smaller than the set addition threshold, adding the p value minimum item of the F statistic in all the potential items meeting the requirements into the current stepwise regression equation, and repeating the steps; otherwise, jumping to S23;
s23, if some potential item added into the model exists, and the p value of the F statistic is larger than the set deletion threshold, removing the p value maximum item of the F statistic in all the potential items meeting the requirements from the current stepwise regression equation, and jumping to S22; otherwise, ending;
s24, obtaining a stepwise regression equation of the current moment;
and S25, repeating the steps S1-S4, and constructing a stepwise regression equation of each moment except the current moment.
Preferably, the merging the stepwise regression equations at all the time points by using a weighted moving average method based on the time series characteristics of the original body measurement data set, and obtaining the final weighted regression equation includes:
s31, acquiring a regression equation matrix at each moment based on the stepwise regression equation at each moment of the original body measurement data set;
s32, constructing a linear regression model at each moment based on the regression equation matrix at each moment of the original body measurement data set;
and S33, merging the linear regression models of the original body measurement data set at each moment by using a weighted moving average method to obtain a final weighted regression equation.
In a second aspect, the present invention further provides a dynamic health evaluation system based on time series physical measurement data, the system includes:
the data set differentiation module is used for acquiring an evaluation index attribute set based on the acquired original physical measurement data set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation among evaluation index attributes; differentiating and classifying the raw physical measurement dataset into a plurality of data subsets based on the evaluation index attribute subsets;
the step-by-step regression equation building module is used for building a step-by-step regression equation based on the evaluation index attribute subset and the data subset corresponding to each moment of the original physical measurement data set;
the weighted regression equation building module is used for merging stepwise regression equations at all moments by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set to obtain a final weighted regression equation;
and the dynamic health evaluation module is used for carrying out dynamic health evaluation by utilizing a final weighted regression equation based on the time sequence body measurement data to be evaluated.
Preferably, the data set differentiation module acquires an evaluation index attribute set based on the acquired original physical measurement data set, and divides the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation among evaluation index attributes; differentiating and classifying the raw fitness dataset into a number of subsets of data based on the subset of evaluation index attributes comprises:
s11, determining the evaluation index attribute of the original body measurement data required by the construction of the stepwise regression equation based on the distribution characteristics and the correlation of the original body measurement data set, forming an evaluation index attribute set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets according to the strong correlation among the evaluation index attributes;
s12, calculating the quantitative result of the evaluation index attribute in the evaluation index attribute subset;
and S13, classifying the original body measurement data set based on the quantification result to obtain a plurality of data subsets.
Preferably, the evaluation index attributes include height and weight;
the quantitative result of the evaluation index attribute includes a BMI index:
wherein the content of the first and second substances,
represents body weight; h represents height.
Preferably, the constructing a stepwise regression equation based on the evaluation index attribute subset and the data subset corresponding to each time of the original physical measurement data set includes:
and constructing a stepwise regression equation by taking the evaluation index attribute subset at each moment as a potential item, wherein the process of constructing the stepwise regression equation specifically comprises the following steps:
s21, constructing a stepwise regression initial equation at the current moment;
s22, if a certain potential item which is not added into the model exists and the p value of the F statistic is smaller than the set addition threshold, adding the p value minimum item of the F statistic in all the potential items meeting the requirements into the current stepwise regression equation, and repeating the steps; otherwise, jumping to S23;
s23, if some potential item added into the model exists, and the p value of the F statistic is larger than the set deletion threshold, removing the p value maximum item of the F statistic in all the potential items meeting the requirements from the current stepwise regression equation, and jumping to S22; otherwise, ending;
s24, obtaining a stepwise regression equation of the current moment;
and S25, repeating the steps S1-S4, and constructing a stepwise regression equation of each moment except the current moment.
Preferably, the weighted regression equation building module merges stepwise regression equations at all times by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set, and the obtaining of the final weighted regression equation includes:
s31, acquiring a regression equation matrix at each moment based on the stepwise regression equation at each moment of the original body measurement data set;
s32, constructing a linear regression model at each moment based on the regression equation matrix at each moment of the original body measurement data set;
and S33, merging the linear regression models of the original body measurement data set at each moment by using a weighted moving average method to obtain a final weighted regression equation.
(III) advantageous effects
The invention provides a dynamic health evaluation method and system based on time sequence body measurement data. Compared with the prior art, the method has the following beneficial effects:
1. firstly, determining an evaluation index attribute set of time sequence physical measurement data, acquiring a division data set based on the evaluation index attribute set, and then constructing a stepwise regression equation based on an evaluation index attribute subset and a data subset at each moment; and finally merging the stepwise regression equations of all the moments of the data set by using a weighted moving average method based on the time sequence characteristics to obtain a final weighted regression equation, and performing dynamic health evaluation by using the final weighted regression equation based on the time sequence body measurement data. The method is based on extensive medical physical examination data to establish the evaluation index attribute set and classify the data set in a dividing way, and the establishment of the evaluation index attribute set and the classification of the data set in a dividing way are screened and reused by a stepwise regression algorithm, so that the usability and the universality of the model are obviously improved, and meanwhile, compared with the traditional model, the precision of the evaluation model is improved;
2. the evaluation index attribute set of the time sequence physical measurement data is determined, the time sequence physical measurement data set is differentiated and classified based on the attribute set, and then a dynamic health evaluation model suitable for different crowds and different characteristics is established based on the evaluation index attribute set and the data set, so that the sample set is segmented and integrated in a multi-level manner, and the dynamic health evaluation of the model realizes the flexibility and the robustness which are changed according to the characteristics of the crowds and the characteristics of the sample;
3. compared with the traditional health evaluation model, the time sequence characteristic of time sequence body measurement data is fused, the influence of time sequence factors on the evaluation result is fully considered on the basis of classical fitting and evaluation, and the characteristic of a time sequence body side data set is fused on different dimensions in a weighted sliding average mode, so that more comprehensive dynamic characteristics are added to the model by using the time sequence characteristic on the basis of not losing the advantages of the original evaluation method, the real situation of the time sequence body measurement data is better met, and the evaluation result is more accurate.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flowchart of a dynamic health assessment method based on time series physical measurements according to an embodiment of the present disclosure;
fig. 2 is a schematic diagram of health index segmentation based on BMI index in the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described, and it is obvious that the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the application provides a dynamic health evaluation method and system based on time sequence body measurement data, solves the problems of inaccurate evaluation result and poor generalization of the existing health evaluation method, and realizes the generalization and robustness of the health evaluation method which are changed according to the characteristics of the crowd and the characteristics of a sample.
In order to solve the technical problems, the general idea of the embodiment of the application is as follows:
in order to solve the problem that the existing health evaluation technology only can analyze a certain current specific medical scene or specific disease and can not screen and multiplex different characteristics due to subjective and single physical measurement data information source and solve the problem that the existing health evaluation technology can not screen and multiplex different characteristics, the invention firstly determines an evaluation index attribute set of an original physical measurement data set and differentiates and classifies the original physical measurement data set based on the evaluation index attribute set to obtain a data subset; then, a stepwise regression equation is constructed for the evaluation index attribute subset and the data subset at each moment; and finally merging the stepwise regression equations of the original body measurement data set at all the moments by using a weighted moving average method based on the time sequence characteristics of the original body measurement data to obtain a final weighted regression equation. Thus, dynamic health assessment can be performed on different patients based on time series physical measurement data to be evaluated by using the final weighted regression equation.
In order to better understand the technical solution, the technical solution will be described in detail with reference to the drawings and the specific embodiments.
Example 1:
in a first aspect, the present invention first provides a dynamic health evaluation method based on time series physical measurement data, referring to fig. 1, the method includes:
s1, acquiring an evaluation index attribute set based on the acquired original body measurement data set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation among evaluation index attributes; differentiating and classifying the raw physical measurement dataset into a plurality of data subsets based on the evaluation index attribute subsets;
s2, constructing a stepwise regression equation based on the evaluation index attribute subset and the data subset corresponding to each moment of the original physical measurement data set;
s3, merging the stepwise regression equations at all the moments by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set to obtain a final weighted regression equation;
and S4, performing dynamic health evaluation by using the final weighted regression equation based on the time sequence body measurement data to be evaluated.
According to the method, firstly, an evaluation index attribute set of time sequence physical measurement data is determined, a division data set is obtained based on the evaluation index attribute set, and then a stepwise regression equation is constructed based on an evaluation index attribute subset and a data subset at each moment; and finally merging the stepwise regression equations of all the moments of the data set by using a weighted moving average method based on the time sequence characteristics to obtain a final weighted regression equation, and performing dynamic health evaluation by using the final weighted regression equation based on the time sequence body measurement data. The method is used for establishing the evaluation index attribute set and classifying the data set based on wide medical physical examination data, and screening and multiplexing the establishment of the evaluation index attribute set and the classification of the data set by a stepwise regression algorithm, so that the usability and the universality of the model are remarkably improved, and the precision of the evaluation model is improved compared with the traditional model.
The following describes in detail the implementation of an embodiment of the present invention by taking height and weight as the evaluation indexes in the attribute set of the evaluation indexes and combining the explanation of specific steps S1-S3.
S1, acquiring an evaluation index attribute set based on the acquired original body measurement data set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation among evaluation index attributes; differentiating and classifying the raw fitness dataset into a number of subsets of data based on the evaluation index attribute subsets.
And acquiring original body measurement data, and selecting a part of the original body measurement data as a sample to form an original body measurement data set. The data set contains the physical examination items and physical examination data values of the original physical examination data, and related time sequence information and the like. According to the method, for an original body measurement data dataset, firstly, an evaluation index attribute set of the original body measurement data dataset is determined (the evaluation index is used as a regression index for subsequently constructing a stepwise regression model), and then, based on the determined evaluation index attribute set, the original body measurement data dataset is classified in a dividing manner by combining the distribution characteristics and the correlation of different body measurement data evaluation indexes, so that a divided data subset is obtained. The method comprises the following specific steps:
s11, determining the evaluation index attributes of the original body measurement data required by the construction of the stepwise regression equation based on the distribution characteristics and the correlation of the original body measurement data set, forming an evaluation index attribute set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets according to the strong correlation among the evaluation index attributes.
The evaluation indexes are various and can be determined according to actual needs, for example, height, weight, blood sugar, blood pressure, blood fat, heartbeat, pulse and the like can be used as the evaluation indexes. Particularly, according to strong correlation possibly existing in each regression independent variable and according to the judgment criterion of the relevant medical field, a complete division of the evaluation index attribute set S is carried out in the initial stage of regression data selection, namely S is divided into n evaluation index attribute subsets, namely { S }1,S2,...,Sn}. The evaluation index attribute subset is a plurality of subsets obtained by dividing evaluation index attributes according to the medical field discrimination of each measurement item. The n subsets of evaluation index attributes must simultaneously satisfy the following conditions:
In this embodiment, the height and the weight are taken as two evaluation index attributes, and the two evaluation index attributes form an evaluation index attribute subset, which is defined as SjThen S isj{ height, weight }, and
and importing a corresponding data subset vector according to the set evaluation index attribute subset, wherein the data subset vector is a data sequence in a certain specified sequence according to the set evaluation index attribute subset.
And S12, calculating the quantitative result of the evaluation index attribute in the evaluation index attribute subset. Calculating an evaluation index attribute subset SjAnd (5) evaluating the quantization result corresponding to the index attribute. In this embodiment, height and weight are calculatedAnd evaluating the quantization result corresponding to the index attribute. Specifically, the calculation of the corresponding BMI index is performed based on the height and weight attributes:
wherein the content of the first and second substances,
body weight values in kilograms; h is the height value in meters.
And S13, classifying the original body measurement data set based on the quantification result to obtain a plurality of data subsets.
In the present embodiment, the data set is classified based on the BMI calculation result. Specifically, referring to fig. 2, the data set may be divided into four types of "too low weight", "normal weight", "overweight" and "obesity" according to the BMI index, and the corresponding data set is respectively denoted as N1，N2，N3，N4。
Each evaluation index attribute in the data set is used as a regression independent variable, and a quantization result corresponding to the evaluation index attribute is used as a regression dependent variable and is used as source data of the following regression analysis. In this example, the height and weight were used as regression independent variables, and the BMI value obtained by calculation was used as a regression dependent variable.
In addition, for the data set, each evaluation index attribute subset of the data set at k time points (that is, k time points are sequentially selected in the past with the current investigation time point k as a starting point (including the current time point k)) is respectively obtained, and k groups are selected from the corresponding data, and are marked as { T [ ({ T) ]1,T2,...,Tk}。
And S2, constructing a stepwise regression equation based on the evaluation index attribute subset and the data subset corresponding to each moment of the original physical measurement data set.
When calculating the stepwise regression equation corresponding to a certain moment in the data set, the differentiation data subset and the corresponding evaluation index attribute subset at the moment are utilized to gradually calculate the regression equationThe regression method establishes a stepwise regression equation with a single subset of differentiation data and a single subset of evaluation index attributes as units. Specifically, taking the construction of the stepwise regression equation at time i as an example, for the data subset N at time ii(i ∈ {1,2,3,4}) and the evaluation index attribute subset S corresponding to the time ij(j ∈ {1, 2.,. n }) a stepwise regression equation is constructed.
Stepwise regression algorithm is a systematic method of adding and deleting terms in a multi-linear model based on statistical significance in the regression. The method starts with an initial model and then compares the interpretability of progressively larger and smaller models, respectively, and updates and decisions of the models are made based thereon. At each step of the update, the p-value of the F statistic is computed, thereby examining the possible effect of the current potential term on the model. If the current potential item has not been added to the model, the following original assumptions are first set:
the corresponding coefficient after the potential term is added to the current model is zero. Then if there is sufficient evidence to reject the original hypothesis, add the term to the current model;
otherwise, if the current potential term already exists in the current model, the corresponding original hypothesis has coefficients of zero, and then if there is not sufficient evidence to reject the original hypothesis, the potential term is removed from the model.
Specifically, the algorithm is explained as follows:
algorithm 1 stepwise regression model
I time of Input data subset NiAnd a current evaluation index attribute subset Sj；
Result stepwise regression equation
Wherein s isjSubset S of attributes representing evaluation indexjBase number of and sj＝|Sj|，αkA weight representing a kth evaluation index attribute in the evaluation index attribute subset; sj[k]Subset S of attributes representing evaluation indexjThe kth element of (1); y isijPresentation evaluationAnd (4) total record of each evaluation index attribute weight in the index attribute subset.
The stepwise regression equation is constructed as follows:
and S21, constructing a stepwise regression initial equation of the current moment.
S22, if some potential item which is not added into the model exists and the p value of the F statistic is smaller than the set adding threshold (namely if the potential item is added into the model, enough evidence is provided that the corresponding coefficient is not zero), adding the p value minimum item of the F statistic in all the potential items which meet the requirement into the current stepwise regression equation, and repeating the steps; otherwise, S23.
S23, if some potential item added into the model exists, the p value of the F statistic is larger than the set deletion threshold (namely, sufficient evidence does not exist to deny the assumption that the corresponding coefficient is zero), removing the p value maximum item of the F statistic in all the potential items meeting the requirement from the current stepwise regression equation, S22; otherwise, ending.
And S24, obtaining a stepwise regression equation of the current moment.
And S25, repeating the steps S1-S4, and constructing a stepwise regression equation of each moment except the current moment.
It is worth noting that the method can build different models from the same set of potential terms, depending on the terms contained in the initial model and the order in which the terms are moved in and out. The method terminates when there is no single step to improve the model. However, there is no guarantee that a different initial model or a different sequence of steps will not result in a better match. In this sense, the stepwise model is locally optimal, but may not be globally optimal.
And S3, merging the stepwise regression equations at all the moments by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set to obtain a final weighted regression equation.
And S31, acquiring a regression equation matrix at each moment based on the stepwise regression equation at each moment of the original body measurement data set.
The stepwise regression equations corresponding to all the moments of the original body measurement data set form a stepwise regression equation set, and a stepwise regression equation matrix can be constructed on the basis of the stepwise regression equation set.
The construction of the stepwise regression matrix and the solving process of the linear regression model are as follows:
based on the above stepwise regression equations, we can obtain the corresponding regression equation matrix M:
s32, constructing a linear regression model at each moment based on the regression equation matrix at each moment of the original body measurement data set.
Based on the regression matrix, a linear regression model is constructed as follows:
y＝Γ*[β1β2β3β4]T*MX
wherein X is a regression independent variable complete set; gamma is a parameter matrix of linear regression; t represents the transpose of the matrix; beta is ai(i e {1,2,3,4}) is a characterization factor for characterizing whether X belongs to the data subset N at time ii，βiThe following constraints are satisfied:
and MXCalculated from the input independent variable X via the regression equation matrix M. Further, element-by-element matrix multiplication is set as a calculation scheme of the above regression equation, and the element-by-element matrix multiplication is formulated as follows:
in the parameter solving process, various parameters in the gamma are regressed by adopting a gradient descent method. The method comprises the following specific steps:
1) selecting gamma0As an initial value, the gradient of the loss function F at that point is calculated
The function F is known to be in the opposite direction of the gradient at that point
The decrease is fastest;
2) if it is selected
For real values where γ > 0 is sufficiently small, then F (Γ)1)＜F(Γ0)；
3) Repeating the descending step, and adjusting the iteration step size to enable the sequence { gammaiConverge to the corresponding extremum.
Calculating corresponding linear regression models according to the stepwise regression equation matrix, and respectively aligning T according to the method1,T2,...,TkEstablishing a linear regression model, and respectively recording as F1,F2,...,Fk。
And S33, merging the linear regression models of the original body measurement data set at each moment by using a weighted moving average method and based on the time sequence characteristics of the data to obtain a final weighted regression equation. The method comprises the following specific steps:
and combining the linear regression models obtained at different moments of the data set by using a weighted moving average method to obtain a final weighted regression equation. When the weighted moving average method is used for setting the relevant weight corresponding to the regression equation at each moment, a weight distribution formula with the weight gradually and slowly reduced along with the time lapse is sequentially established by adopting the principle that the closer the weight is to the current moment, the larger the weight is. Meanwhile, the weight distribution fully adopts and respects the expert opinions, and the weight is properly adjusted, so that the result objectivity is ensured.
Based on the linear regression model set, a weighted moving average equation is established:
F＝w1F1+w2F2+...+wkFk
wherein, the calculation and normalization processing of the weight in the moving average equation are performed according to the following formula.
wiRepresenting the assigned weights of the ith linear regression equation.
The weight proportion of the current weighted moving average equation meets the exponential decay condition, and a linear decreasing decay strategy is adopted for different time periods and distribution characteristics of corresponding data sets in practical application. Specifically, after obtaining the time distribution of the known body measurement features, comparative evaluation can be performed through different weight calculation modes as described below. The applicable weight distribution in this solution is shown in table 1 below.
Based on this, a weighted moving average equation in the form of a corresponding regression can be established, wherein the training pattern of the hyper-parameters can be processed with reference to the gradient descent method described above. And evaluating each average equation by adopting an expected risk minimization principle so as to establish a final weighted regression equation.
And S4, performing dynamic health evaluation by using the final weighted regression equation based on the time sequence body measurement data to be evaluated.
Aiming at different characteristic crowds, unique time sequence body measurement data of the crowds are used, different index evaluation attributes are used, and finally obtained weighted regression equations are used for dynamic health assessment.
Thus, the whole process of the dynamic health evaluation method based on the time sequence physical measurement data is completed.
Example 2:
in a second aspect, the present invention further provides a dynamic health evaluation system based on time series physical measurement data, the system comprising:
the data set differentiation module is used for acquiring an evaluation index attribute set based on the acquired original physical measurement data set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation among evaluation index attributes; differentiating and classifying the raw physical measurement dataset into a plurality of data subsets based on the evaluation index attribute subsets;
the step-by-step regression equation building module is used for building a step-by-step regression equation based on the evaluation index attribute subset and the data subset corresponding to each moment of the original physical measurement data set;
the weighted regression equation building module is used for merging stepwise regression equations at all moments by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set to obtain a final weighted regression equation;
and the dynamic health evaluation module is used for carrying out dynamic health evaluation by utilizing a final weighted regression equation based on the time sequence body measurement data to be evaluated.
Optionally, the data set differentiation module acquires an evaluation index attribute set based on the acquired original physical measurement data set, and divides the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation between evaluation index attributes; differentiating and classifying the raw fitness dataset into a number of subsets of data based on the subset of evaluation index attributes comprises:
s11, determining the evaluation index attribute of the original body measurement data required by the construction of the stepwise regression equation based on the distribution characteristics and the correlation of the original body measurement data set, forming an evaluation index attribute set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets according to the strong correlation among the evaluation index attributes;
s12, calculating the quantitative result of the evaluation index attribute in the evaluation index attribute subset;
and S13, classifying the original body measurement data set based on the quantification result to obtain a plurality of data subsets.
Preferably, the evaluation index attributes include height and weight;
the quantitative result of the evaluation index attribute includes a BMI index:
wherein the content of the first and second substances,
represents body weight; h represents height.
Optionally, the constructing a stepwise regression equation based on the evaluation index attribute subset and the data subset corresponding to each time of the original physical measurement data set includes:
and constructing a stepwise regression equation by taking the evaluation index attribute subset at each moment as a potential item, wherein the process of constructing the stepwise regression equation specifically comprises the following steps:
s21, constructing a stepwise regression initial equation at the current moment;
s22, if a certain potential item which is not added into the model exists and the p value of the F statistic is smaller than the set addition threshold, adding the p value minimum item of the F statistic in all the potential items meeting the requirements into the current stepwise regression equation, and repeating the steps; otherwise, jumping to S23;
s23, if some potential item added into the model exists, and the p value of the F statistic is larger than the set deletion threshold, removing the p value maximum item of the F statistic in all the potential items meeting the requirements from the current stepwise regression equation, and jumping to S22; otherwise, ending;
s24, obtaining a stepwise regression equation of the current moment;
and S25, repeating the steps S1-S4, and constructing a stepwise regression equation of each moment except the current moment.
Optionally, the weighted regression equation building module merges stepwise regression equations at all times by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set, and acquiring a final weighted regression equation includes:
s31, acquiring a regression equation matrix at each moment based on the stepwise regression equation at each moment of the original body measurement data set;
s32, constructing a linear regression model at each moment based on the regression equation matrix at each moment of the original body measurement data set;
and S33, merging the linear regression models of the original body measurement data set at each moment by using a weighted moving average method to obtain a final weighted regression equation.
It can be understood that the dynamic health evaluation system based on time series physical measurement data provided in the embodiment of the present invention corresponds to the above dynamic health evaluation method based on time series physical measurement data, and the explanation, examples, and beneficial effects of the relevant contents thereof may refer to the corresponding contents in the dynamic health evaluation method based on time series physical measurement data, and are not described herein again.
In summary, compared with the prior art, the method has the following beneficial effects:
1. firstly, determining an evaluation index attribute set of time sequence physical measurement data, acquiring a division data set based on the evaluation index attribute set, and then constructing a stepwise regression equation based on an evaluation index attribute subset and a data subset at each moment; and finally merging the stepwise regression equations of all the moments of the data set by using a weighted moving average method based on the time sequence characteristics to obtain a final weighted regression equation, and performing dynamic health evaluation by using the final weighted regression equation based on the time sequence body measurement data. The method is based on extensive medical physical examination data to establish the evaluation index attribute set and classify the data set in a dividing way, and the establishment of the evaluation index attribute set and the classification of the data set in a dividing way are screened and reused by a stepwise regression algorithm, so that the usability and the universality of the model are obviously improved, and meanwhile, compared with the traditional model, the precision of the evaluation model is improved;
2. the evaluation index attribute set of the time sequence physical measurement data is determined, the time sequence physical measurement data set is differentiated and classified based on the attribute set, and then a dynamic health evaluation model suitable for different crowds and different characteristics is established based on the evaluation index attribute set and the data set, so that the sample set is segmented and integrated in a multi-level manner, and the dynamic health evaluation of the model realizes the flexibility and the robustness which are changed according to the characteristics of the crowds and the characteristics of the sample;
3. compared with the traditional health evaluation model, the time sequence characteristic of time sequence body measurement data is fused, the influence of time sequence factors on the evaluation result is fully considered on the basis of classical fitting and evaluation, and the characteristic of a time sequence body side data set is fused on different dimensions in a weighted sliding average mode, so that more comprehensive dynamic characteristics are added to the model by using the time sequence characteristic on the basis of not losing the advantages of the original evaluation method, the real situation of the time sequence body measurement data is better met, and the evaluation result is more accurate.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A dynamic health evaluation method based on time series physical measurement data is characterized by comprising the following steps:
acquiring an evaluation index attribute set based on the acquired original physical measurement data set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation among evaluation index attributes; differentiating and classifying the raw physical measurement dataset into a plurality of data subsets based on the evaluation index attribute subsets;
constructing a stepwise regression equation based on the evaluation index attribute subset and the data subset corresponding to each moment of the original physical measurement data set;
merging stepwise regression equations at all moments by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set to obtain a final weighted regression equation;
and performing dynamic health evaluation by using a final weighted regression equation based on the time sequence body measurement data to be evaluated.
2. The method of claim 1, wherein the determining a set of evaluation index attributes for the time series fitness data and obtaining a partitioned data set based on the set of evaluation index attributes comprises:
s11, determining the evaluation index attribute of the original body measurement data required by the construction of the stepwise regression equation based on the distribution characteristics and the correlation of the original body measurement data set, forming an evaluation index attribute set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets according to the strong correlation among the evaluation index attributes;
s12, calculating the quantitative result of the evaluation index attribute in the evaluation index attribute subset;
and S13, classifying the original body measurement data set based on the quantification result to obtain a plurality of data subsets.
3. The method of claim 2, wherein the evaluation index attributes comprise height and weight;
the quantitative result of the evaluation index attribute includes a BMI index:
wherein the content of the first and second substances,
represents body weight;hindicating height.
4. The method of claim 1, wherein constructing a stepwise regression equation based on the subset of evaluation index attributes and the data subset corresponding to each time instant of the raw fitness measure dataset comprises:
and constructing a stepwise regression equation by taking the evaluation index attribute subset at each moment as a potential item, wherein the process of constructing the stepwise regression equation specifically comprises the following steps:
s21, constructing a stepwise regression initial equation at the current moment;
s22, if a certain potential item which is not added into the model exists and the p value of the F statistic is smaller than the set addition threshold, adding the p value minimum item of the F statistic in all the potential items meeting the requirements into the current stepwise regression equation, and repeating the steps; otherwise, jumping to S23;
s23, if some potential item added into the model exists, and the p value of the F statistic is larger than the set deletion threshold, removing the p value maximum item of the F statistic in all the potential items meeting the requirements from the current stepwise regression equation, and jumping to S22; otherwise, ending;
s24, obtaining a stepwise regression equation of the current moment;
and S25, repeating the steps S1-S4, and constructing a stepwise regression equation of each moment except the current moment.
5. The method of claim 1, wherein merging stepwise regression equations at all times based on time series characteristics of the raw anthropometric dataset using a weighted moving average method, and obtaining a final weighted regression equation comprises:
s31, acquiring a regression equation matrix at each moment based on the stepwise regression equation at each moment of the original body measurement data set;
s32, constructing a linear regression model at each moment based on the regression equation matrix at each moment of the original body measurement data set;
and S33, merging the linear regression models of the original body measurement data set at each moment by using a weighted moving average method to obtain a final weighted regression equation.
6. A dynamic health assessment system based on time series anthropometric data, the system comprising:
the data set differentiation module is used for acquiring an evaluation index attribute set based on the acquired original physical measurement data set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets based on strong correlation among evaluation index attributes; differentiating and classifying the raw physical measurement dataset into a plurality of data subsets based on the evaluation index attribute subsets;
the step-by-step regression equation building module is used for building a step-by-step regression equation based on the evaluation index attribute subset and the data subset corresponding to each moment of the original physical measurement data set;
the weighted regression equation building module is used for merging stepwise regression equations at all moments by using a weighted moving average method based on the time sequence characteristics of the original body measurement data set to obtain a final weighted regression equation;
and the dynamic health evaluation module is used for carrying out dynamic health evaluation by utilizing a final weighted regression equation based on the time sequence body measurement data to be evaluated.
7. The system of claim 6, wherein the dataset differentiation module obtains a set of evaluation index attributes based on the obtained raw fitness dataset, the set of evaluation index attributes being divided into a number of evaluation index attribute subsets based on strong correlations between evaluation index attributes; differentiating and classifying the raw fitness dataset into a number of subsets of data based on the subset of evaluation index attributes comprises:
s11, determining the evaluation index attribute of the original body measurement data required by the construction of the stepwise regression equation based on the distribution characteristics and the correlation of the original body measurement data set, forming an evaluation index attribute set, and dividing the evaluation index attribute set into a plurality of evaluation index attribute subsets according to the strong correlation among the evaluation index attributes;
s12, calculating the quantitative result of the evaluation index attribute in the evaluation index attribute subset;
and S13, classifying the original body measurement data set based on the quantification result to obtain a plurality of data subsets.
8. The system of claim 7, wherein the evaluation index attributes include height and weight;
the quantitative result of the evaluation index attribute includes a BMI index:
wherein the content of the first and second substances,
represents body weight; h represents height.
9. The system of claim 6, wherein the constructing a stepwise regression equation based on the subset of evaluation index attributes and the subset of data corresponding to each time instance of the raw fitness measure dataset comprises:
and constructing a stepwise regression equation by taking the evaluation index attribute subset at each moment as a potential item, wherein the process of constructing the stepwise regression equation specifically comprises the following steps:
s21, constructing a stepwise regression initial equation at the current moment;
s22, if a certain potential item which is not added into the model exists and the p value of the F statistic is smaller than the set addition threshold, adding the p value minimum item of the F statistic in all the potential items meeting the requirements into the current stepwise regression equation, and repeating the steps; otherwise, jumping to S23;
s23, if some potential item added into the model exists, and the p value of the F statistic is larger than the set deletion threshold, removing the p value maximum item of the F statistic in all the potential items meeting the requirements from the current stepwise regression equation, and jumping to S22; otherwise, ending;
s24, obtaining a stepwise regression equation of the current moment;
and S25, repeating the steps S1-S4, and constructing a stepwise regression equation of each moment except the current moment.
10. The system of claim 6, wherein the weighted regression equation building module merges the stepwise regression equations at all times using a weighted moving average based on the time series characteristics of the raw anthropometric dataset, and obtaining the final weighted regression equation comprises:
s31, acquiring a regression equation matrix at each moment based on the stepwise regression equation at each moment of the original body measurement data set;
s32, constructing a linear regression model at each moment based on the regression equation matrix at each moment of the original body measurement data set;
and S33, merging the linear regression models of the original body measurement data set at each moment by using a weighted moving average method to obtain a final weighted regression equation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110945416.0A CN113851221A (en) | 2021-08-17 | 2021-08-17 | Dynamic health evaluation method and system based on time sequence body measurement data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110945416.0A CN113851221A (en) | 2021-08-17 | 2021-08-17 | Dynamic health evaluation method and system based on time sequence body measurement data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113851221A true CN113851221A (en) | 2021-12-28 |
Family
ID=78975800
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110945416.0A Pending CN113851221A (en) | 2021-08-17 | 2021-08-17 | Dynamic health evaluation method and system based on time sequence body measurement data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113851221A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115908069A (en) * | 2023-01-05 | 2023-04-04 | 中大体育产业集团股份有限公司 | Intelligent management method and system for body measurement data of primary and middle school students |
-
2021
- 2021-08-17 CN CN202110945416.0A patent/CN113851221A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115908069A (en) * | 2023-01-05 | 2023-04-04 | 中大体育产业集团股份有限公司 | Intelligent management method and system for body measurement data of primary and middle school students |
| CN115908069B (en) * | 2023-01-05 | 2023-09-08 | 中大体育产业集团股份有限公司 | Intelligent management method and system for body measurement data of primary and secondary school students |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111292853B (en) | Multi-parameter-based cardiovascular disease risk prediction network model and construction method thereof | |
| CN111261282A (en) | Sepsis early prediction method based on machine learning | |
| CN109528197B (en) | Individual prediction method and system for mental diseases based on brain function map | |
| CN109920547A (en) | A kind of diabetes prediction model construction method based on electronic health record data mining | |
| CN107169284A (en) | A kind of biomedical determinant attribute system of selection | |
| KR20210114012A (en) | Diagnosis and Effectiveness of Attention Deficit Hyperactivity Disorder Monitoring | |
| Abd El Aal et al. | An optimized RNN-LSTM approach for parkinson’s disease early detection using speech features | |
| CN111631719A (en) | Method for predicting falling risk of old people | |
| CN114943629A (en) | Health management and health care service system and health management method thereof | |
| CN113593708A (en) | Sepsis prognosis prediction method based on integrated learning algorithm | |
| CN113851221A (en) | Dynamic health evaluation method and system based on time sequence body measurement data | |
| CN114628033A (en) | Disease risk prediction method, device, equipment and storage medium | |
| CN117609813B (en) | Intelligent management method for intensive patient monitoring data | |
| Verma et al. | Modelling similarity for comparing physical activity profiles-a data-driven approach | |
| CN108664949A (en) | The method classified to epileptic electroencephalogram (eeg) figure signal using radial base neural net | |
| CN115147768B (en) | Fall risk assessment method and system | |
| CN115527673A (en) | Mental health risk screening system and method based on big data | |
| CN112382382B (en) | Cost-sensitive integrated learning classification method and system | |
| CN108346471A (en) | A kind of analysis method and device of pathological data | |
| CN113868597A (en) | Regression fairness measurement method for age estimation | |
| Rajmohan et al. | G-Sep: A Deep Learning Algorithm for Detection of Long-Term Sepsis Using Bidirectional Gated Recurrent Unit | |
| Malik et al. | Heart Disease Prediction Using Artificial Intelligence | |
| CN113284612B (en) | Survival analysis method based on XGBoost algorithm | |
| CN117598700B (en) | Intelligent blood oxygen saturation detection system and method | |
| CN117079821B (en) | Patient hospitalization event prediction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |