CN-122024975-A - Personalized health management scheme data mining method

CN122024975ACN 122024975 ACN122024975 ACN 122024975ACN-122024975-A

Abstract

The invention relates to the technical field of health management and discloses a personalized health management scheme data mining method which comprises the steps of obtaining multi-source heterogeneous health file data of a target user, carrying out space-time alignment processing on the multi-source heterogeneous health file data to generate a standard health data sequence, constructing a dynamic health entropy value to divide a health grade interval, inputting the health grade interval into a preset health intervention knowledge base, determining intervention measure primitives matched with a health feature vector set, carrying out associated mapping on the intervention measure primitives and the dynamic health entropy value to construct an intervention mapping matrix, carrying out combined screening and priority ordering on the intervention measure primitives in the intervention mapping matrix with the minimized dynamic health entropy value as an optimization target to generate a personalized health candidate set, and carrying out expected effect deduction on the personalized health candidate set to obtain an optimized health management scheme.

Inventors

XIE MINGHUI
WANG XIANJUN
SONG HUI
LIN QIFENG
LIN CHENG

Assignees

福州中康智慧科技股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260410

Claims (10)

1. A personalized health management scheme data mining method, the method comprising: a1, acquiring multi-source heterogeneous health record data of a target user, and performing space-time alignment processing on the multi-source heterogeneous health record data to generate a standard health data sequence of the target user; A2, extracting the physiological index fluctuation characteristics and behavior pattern association characteristics in the standard health data sequence to construct a health characteristic vector set of the target user; A3, carrying out nonlinear aggregation on the deviation degree and the discrete rate of the health feature vector set to obtain a dynamic health entropy value of the target user so as to divide a health grade interval of the target user; A4, inputting the health grade interval into a preset health intervention knowledge base, determining intervention measure primitives matched with the health feature vector set, performing association mapping on the intervention measure primitives and the dynamic health entropy value, and constructing an intervention mapping matrix of the target user; a5, taking the minimized dynamic health entropy value as an optimization target, carrying out combined screening and priority ordering on the intervention measure primitives in the intervention mapping matrix, and generating a personalized health candidate set of the target user; And A6, carrying out expected effect deduction on the personalized health candidate set to obtain an optimized health management scheme of the target user.
2. The method for mining personalized health management scheme data according to claim 1, wherein the obtaining multi-source heterogeneous health record data of the target user, performing space-time alignment processing on the multi-source heterogeneous health record data, and generating the standard health data sequence of the target user comprises: Synchronously collecting wearable equipment monitoring data, electronic medical record data and physical examination report text data of a target user to integrate the data into multi-source heterogeneous health record data of the target user; Taking the timestamp of the monitoring data of the wearable equipment as a reference, and performing linear interpolation alignment on the electronic medical record data and the time tag in the physical examination report text data to obtain synchronous health data of the target user; Based on the synchronous health data, carrying out normalization processing on the electronic medical record data and the physical examination report text data to obtain normalized health data of the target user; And carrying out data classification on the normalized health data, and carrying out difference correction on the missing values in the classified data according to the historical health data average value of the target user to obtain a standard health data sequence of the target user.
3. The method of claim 1, wherein the extracting the physiological index fluctuation feature and behavior pattern association feature in the standard health data sequence to construct the health feature vector set of the target user comprises: Performing data decoupling on the standard health data sequence to obtain a behavior parameter subsequence and a physiological parameter subsequence of the target user; Performing fluctuation analysis on the physiological indexes in the physiological parameter subsequence to obtain the physiological index fluctuation characteristics of the target user; Performing time correlation analysis on the step index and the sleep duration index in the behavior parameter subsequence to obtain behavior pattern correlation characteristics of the target user; And vectorizing and splicing the physiological index fluctuation characteristics and the behavior pattern association characteristics to generate a health characteristic vector set of the target user.
4. The method for mining personalized health management scheme data according to claim 1, wherein the nonlinear aggregation of the deviation degree and the discrete rate of the health feature vector set to obtain the dynamic health entropy value of the target user so as to divide the health class interval of the target user comprises: Extracting the central point of each dimension of feature vector in the health feature vector set, and calculating the mahalanobis distance from each feature vector in the health feature vector set to the central point; Carrying out deviation measurement on the mahalanobis distance to obtain the deviation degree of the health feature vector set; and extracting standard deviation and average value of each dimension of feature vectors in the health feature vector set, and taking the ratio of the standard deviation to the average value as the discrete rate of the health feature vector set.
5. The method of claim 4, wherein said determining the ratio of the standard deviation to the mean as the discrete rate of the health feature vector set comprises: And carrying out nonlinear fusion on the deviation degree and the discrete rate to generate a dynamic health entropy value of the target user, wherein a calculation formula of the dynamic health entropy value is as follows: ; In the formula, For the dynamic health entropy value in question, As a weight coefficient for the degree of deviation, For the degree of deviation to be stated, As a function of the index of the values, As the weight coefficient of the discrete rate, The adjustment coefficient is saturated for a preset degree of deviation, For the said rate of dispersion to be the same, As a logarithmic function; acquiring a preset health grade threshold value set, comparing the dynamic health entropy value with the preset health grade threshold value set, and determining a threshold value interval of the dynamic health entropy value; and according to the threshold interval, matching the health grade label of the preset health grade threshold set, and dividing the health grade interval of the target user.
6. The personalized health management scheme data mining method according to claim 4, wherein said inputting the health class interval into a preset health intervention knowledge base, determining an intervention measure primitive matched with the health feature vector set, comprises: Extracting a grade label corresponding to the health grade interval; Inputting the grade label into the preset health intervention knowledge base, and positioning an intervention strategy index table associated with the grade label; Taking the health feature vector set as a query vector, and performing similarity matching in the intervention strategy index table to obtain a candidate intervention measure set of the target user; extracting the execution conditions and action targets of the intervention measures in the candidate intervention measure set, and eliminating the intervention measures incompatible with the health feature vector set to obtain the intervention measure primitive of the target user.
7. The personalized health management scheme data mining method according to claim 6, wherein the mapping the intervention measure primitive in association with the dynamic health entropy value, constructing an intervention mapping matrix of the target user, comprises: acquiring an action dimension label and an expected adjustment amplitude in the intervention measure primitive; decomposing the dynamic health entropy value to the action dimension label to obtain an entropy component of the dynamic health entropy value; performing product operation on the expected adjusting amplitude and the entropy component to obtain an adjusting intensity value of the intervention measure primitive; and filling the adjustment intensity value to a matrix corresponding position by taking the intervention measure primitive as a row index and the action dimension label as a column index, and constructing an intervention mapping matrix of the target user.
8. The method for mining personalized health management scheme data according to claim 7, wherein said generating a personalized health candidate set for said target user by performing combined screening and prioritization of intervention measure primitives in said intervention map matrix with the minimization of said dynamic health entropy as an optimization objective comprises: according to the adjustment intensity value, identifying a synergistic relationship and a repulsive relationship between the dry pre-measure primitives in the intervention mapping matrix; Performing topology reconstruction on the cooperative relationship and the exclusive relationship to obtain an intervention interaction diagram of the target user; and traversing the intervention measure interaction diagram by taking the minimized dynamic health entropy value as an optimization target, and generating a candidate combination set of the target user.
9. The method for mining personalized health management scheme data according to claim 8, wherein the step of traversing the intervention interaction map with the objective of minimizing the dynamic health entropy value, and generating the candidate combination set of the objective user comprises: according to the predicted descending amplitude of the dynamic health entropy value, descending order arrangement is carried out on the candidate combination sets, and a first ordering sequence of the target user is obtained; acquiring implementation cost labels and implementation risk levels in the candidate combination set; And performing secondary sorting on the candidate combinations in the first sorting sequence according to the implementation cost label and the implementation risk level to obtain the personalized health candidate set of the target user.
10. The method for mining personalized health management scheme data according to claim 2, wherein the performing the expected effect deduction on the personalized health candidate set to obtain the optimized health management scheme of the target user comprises: acquiring an implementation time sequence relation in the personalized health candidate set; Performing time sequence association on the implementation time sequence relation and the historical health data of the target user, and performing path evolution on the associated result to generate an expected health evolution path of the target user; Performing index calibration on the expected healthy evolution path to obtain an effect evaluation index of the target user; and comprehensively evaluating the personalized health candidate set based on the effect evaluation index to obtain an optimized health management scheme of the personalized health candidate set.

Description

Personalized health management scheme data mining method Technical Field The invention relates to the technical field of health management, in particular to a personalized health management scheme data mining method. Background The prior art has obvious defects in the link of health data processing. The method has the advantages that unified space-time alignment and normalization processing cannot be carried out on the multi-source heterogeneous health file data, the data missing value correction mode is simple, a standard health data sequence unified in specification is difficult to form, the physiological index fluctuation characteristic and behavior pattern association characteristic cannot be extracted accurately, a health characteristic vector set which completely reflects the real state of a user cannot be constructed, and the follow-up analysis lacks a reliable data base. The prior art has prominent defects in the links of health assessment and scheme generation. The health state is not quantitatively classified through the dynamic health entropy value, the health grade classification mode is extensive, scientific basis is lacked, accurate matching of intervention measures cannot be realized by combining a health intervention knowledge base, quantitative association is realized by not constructing an intervention mapping matrix, intervention combination screening and priority sorting cannot be performed by using an optimization target, expected effect deduction cannot be performed, and finally the generated health management scheme has poor adaptability and insufficient scientificity, and is difficult to meet personalized health management requirements. In the prior art, although there is an attempt to use mahalanobis distance for medical data analysis, a technical scheme of non-linearly aggregating the deviation degree and the discrete rate of the mahalanobis distance from the health feature vector to construct a dynamic health entropy value is not disclosed. Disclosure of Invention The invention provides a personalized health management scheme data mining method for solving the problems in the background technology. In order to achieve the above object, the present invention provides a personalized health management scheme data mining method, including: a1, acquiring multi-source heterogeneous health record data of a target user, and performing space-time alignment processing on the multi-source heterogeneous health record data to generate a standard health data sequence of the target user; A2, extracting the physiological index fluctuation characteristics and behavior pattern association characteristics in the standard health data sequence to construct a health characteristic vector set of the target user; A3, carrying out nonlinear aggregation on the deviation degree and the discrete rate of the health feature vector set to obtain a dynamic health entropy value of the target user so as to divide a health grade interval of the target user; A4, inputting the health grade interval into a preset health intervention knowledge base, determining intervention measure primitives matched with the health feature vector set, performing association mapping on the intervention measure primitives and the dynamic health entropy value, and constructing an intervention mapping matrix of the target user; a5, taking the minimized dynamic health entropy value as an optimization target, carrying out combined screening and priority ordering on the intervention measure primitives in the intervention mapping matrix, and generating a personalized health candidate set of the target user; And A6, carrying out expected effect deduction on the personalized health candidate set to obtain an optimized health management scheme of the target user. In a preferred embodiment, the obtaining the multi-source heterogeneous health record data of the target user, performing space-time alignment processing on the multi-source heterogeneous health record data, and generating the standard health data sequence of the target user includes: Synchronously collecting wearable equipment monitoring data, electronic medical record data and physical examination report text data of a target user to integrate the data into multi-source heterogeneous health record data of the target user; Taking the timestamp of the monitoring data of the wearable equipment as a reference, and performing linear interpolation alignment on the electronic medical record data and the time tag in the physical examination report text data to obtain synchronous health data of the target user; Based on the synchronous health data, carrying out normalization processing on the electronic medical record data and the physical examination report text data to obtain normalized health data of the target user; And carrying out data classification on the normalized health data, and carrying out difference correction on the missing values in the classified data according to the histori