CN-122025179-A - Important crowd health data analysis method and system based on deep learning

CN122025179ACN 122025179 ACN122025179 ACN 122025179ACN-122025179-A

Abstract

The application relates to the field of health data analysis, and provides a method and a system for analyzing health data of key people based on deep learning, wherein the method comprises the steps of acquiring physiological data of each physiological index in a plurality of physiological indexes acquired by a physiological state acquisition device worn by key people aiming at each key person in the key people; the method comprises the steps of determining whether the physiological data of a target physiological index is abnormal and the abnormal reason of the abnormal target acquisition device according to each target physiological index in a plurality of physiological indexes, adjusting the physiological data of the target physiological index based on the abnormal reason of the target acquisition device of the physiological data of the target physiological index to obtain the adjusted physiological data of the target physiological index, inputting the adjusted physiological data of the plurality of physiological indexes into a preset deep learning model, and obtaining the health analysis result of key personnel output by the preset deep learning model. The accuracy of the health data analysis can be improved.

Inventors

LIU RONGGUAN
SHANG QINGDONG
CHEN PEIJIANG

Assignees

临沂亿通软件有限公司
临沂大学

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. The key crowd health data analysis method based on deep learning is characterized by comprising the following steps of: aiming at each key person in key people, acquiring physiological data of each physiological index in a plurality of physiological indexes acquired by a physiological state acquisition device worn by the key person; Determining whether the physiological data of the target physiological index is abnormal and the abnormal reason of the abnormal target acquisition device according to each target physiological index in the multiple physiological indexes, wherein the target physiological index is any one physiological index in the multiple physiological indexes; Adjusting the physiological data of the target physiological index based on the abnormal reason of the target acquisition device of the physiological data of the target physiological index to obtain the adjusted physiological data of the target physiological index; and inputting the physiological data of the adjustment of the multiple physiological indexes into a preset deep learning model to obtain the health analysis result of the key personnel output by the preset deep learning model.
2. The method for analyzing health data of a key crowd based on deep learning according to claim 1, wherein determining whether abnormality occurs in physiological data of the target physiological index and an abnormality cause of an abnormal target acquisition device comprises: The physiological data characteristic information of each preset abnormal reason in a plurality of preset abnormal reasons of the target physiological index is obtained, wherein the physiological data characteristic information comprises physiological data statistical characteristics, physiological data frequency domain characteristics and physiological data time domain characteristics; determining a first similarity between physiological data of the target physiological index and physiological data characteristic information of a preset abnormality cause for each of a plurality of preset abnormality causes; and determining whether the physiological data of the target physiological index is abnormal or not and the abnormal reasons of the abnormal target acquisition device according to the first similarity of the preset abnormal reasons.
3. The method for analyzing health data of a key crowd based on deep learning according to claim 2, wherein determining whether abnormality occurs in physiological data of the target physiological index and an abnormality cause of an abnormal target acquisition device according to first similarities of a plurality of preset abnormality causes comprises: Determining that the physiological data of the target physiological index is abnormal when the largest first similarity among the first similarities of the preset abnormal reasons is larger than a preset similarity threshold; Taking a preset abnormal reason with the first similarity larger than the preset similarity threshold value as an initial abnormal reason; the method comprises the steps of obtaining a first information set of initial abnormality reasons according to each initial abnormality reason, wherein the first information set comprises a plurality of pieces of first information, wherein the first information comprises physiological data characteristic information of an associated physiological index associated with a target physiological index in a plurality of physiological indexes when the target physiological index is abnormal due to the initial abnormality reasons; Determining, for each first information in the first information set, a second similarity of physiological data of the associated physiological index to the first information; Taking the weighted sum of the second similarity as a third similarity of the initial abnormality cause; Taking a weighted sum of the third similarity and the first similarity as a fourth similarity of the initial abnormality cause; and taking the initial abnormality reason with the fourth maximum similarity as the abnormality reason of the target acquisition device.
4. The method for analyzing health data of a key crowd based on deep learning according to claim 1, wherein when a sensor of the target physiological index collecting device is aged due to abnormality of the target collecting device, adjusting physiological data of the target physiological index based on the abnormality of the target collecting device of the physiological data of the target physiological index to obtain adjusted physiological data of the target physiological index, comprises: acquiring the total working time length of a sensor of the physiological state acquisition device; And adjusting the amplitude of the physiological data of the target physiological index according to the total working duration to obtain the adjusted physiological data of the target physiological index.
5. The method for analyzing health data of key crowd based on deep learning according to claim 4, wherein adjusting the amplitude of the physiological data of the target physiological index according to the total working time length to obtain the adjusted physiological data of the target physiological index comprises: the method comprises the steps of obtaining a first corresponding relation, wherein the first corresponding relation comprises a one-to-one corresponding relation between a plurality of working time ranges and a plurality of amplitude adjustment coefficients; taking an amplitude adjustment coefficient corresponding to a working duration range in which the working total duration is located in the first corresponding relation as a target amplitude adjustment coefficient; Taking the product of the amplitude of the physiological data of the target physiological index and the target amplitude adjustment coefficient as the amplitude of the physiological data of the target physiological index after adjustment, and obtaining the adjusted physiological data of the target physiological index.
6. The method for analyzing health data of key people based on deep learning according to claim 5, wherein the health analysis result includes health state information and a confidence level of the health state information, and after obtaining the health analysis result of the key people output by the preset deep learning model, the method further includes: when the confidence coefficient is smaller than a preset confidence coefficient threshold value, increasing each amplitude adjustment coefficient in the first corresponding relation by a preset adjustment step length to obtain a second corresponding relation; Re-determining the health analysis result of the key personnel according to the second corresponding relation; reducing each amplitude adjustment coefficient in the first corresponding relation by the preset adjustment step length to obtain a third corresponding relation; re-determining the health analysis result of the key personnel according to the third corresponding relation; and taking the corresponding relation corresponding to the maximum value in the confidence coefficient of the two newly determined health analysis results as the adjusted first corresponding relation.
7. The method for analyzing health data of a key crowd based on deep learning according to claim 1, wherein the step of acquiring the physiological data of each of the plurality of physiological indexes acquired by the physiological state acquisition device worn by the key crowd comprises the steps of: acquiring the original data of each physiological index in a plurality of physiological indexes acquired by a physiological state acquisition device worn by the key personnel; And carrying out outlier processing on the original data of each physiological index to obtain the physiological data of each physiological index.
8. The method for analyzing health data of a key crowd based on deep learning according to claim 1, wherein when the reason for abnormality of the target acquisition device is abnormal in the wearing mode of the physiological state acquisition device, adjusting the physiological data of the target physiological index based on the reason for abnormality of the target acquisition device of the physiological data of the target physiological index to obtain adjusted physiological data of the target physiological index comprises: dividing the physiological data of the target physiological index into normal physiological data and artifact physiological data according to a preset independent component analysis model or a preset empirical mode decomposition model; And taking the normal physiological data as the adjustment physiological data of the target physiological index.
9. The method for analyzing health data of key crowd based on deep learning according to claim 8, wherein after obtaining the adjusted physiological data of the target physiological index, the method further comprises: The equipment adjustment indication information is used for indicating and adjusting the wearing mode of the physiological state acquisition device on the key personnel; And sending the device adjustment indication information to target equipment so that the target equipment displays the device adjustment indication information.
10. The key crowd health data analysis system based on deep learning is characterized by comprising an acquisition device and a processing device; The acquisition device is used for acquiring the physiological data of each physiological index in the plurality of physiological indexes acquired by the physiological state acquisition device worn by the key people aiming at each key person in the key crowd; The processing device is used for determining whether the physiological data of the target physiological index is abnormal and the abnormal reason of the abnormal target acquisition device according to each target physiological index in the multiple physiological indexes, wherein the target physiological index is any physiological index in the multiple physiological indexes; The processing device is used for adjusting the physiological data of the target physiological index based on the abnormal reason of the target acquisition device of the physiological data of the target physiological index to obtain the adjusted physiological data of the target physiological index; the processing device is used for inputting the physiological data of the adjustment of the multiple physiological indexes into a preset deep learning model to obtain the health analysis result of the key personnel output by the preset deep learning model.

Description

Important crowd health data analysis method and system based on deep learning Technical Field The application relates to the field of health data analysis, in particular to a method and a system for analyzing key crowd health data based on deep learning. Background With the increasing popularity of medical health data collection devices, medical institutions face significant challenges in managing and utilizing vast amounts, sources, and different formats of health data. When the traditional processing method is used for analyzing the health data based on the detected physiological data, whether the physiological data are normal or not is generally judged, then the normal physiological data are reserved, abnormal physiological data are discarded, and then the health data analysis is carried out, however, some abnormal physiological data reflect the health state, and the discarding of the abnormal physiological data tends to reduce the accuracy of the health data analysis. Disclosure of Invention The application provides a key crowd health data analysis method and system based on deep learning, which can improve the accuracy of health data analysis. In order to achieve the above purpose, the application adopts the following technical scheme: The application discloses a key crowd health data analysis method based on deep learning, which comprises the steps of acquiring physiological data of each physiological index in a plurality of physiological indexes collected by a physiological state collection device worn by key people for each key person in the key crowd, determining whether the physiological data of the target physiological index is abnormal and the abnormal reason of the abnormal target collection device for each target physiological index in the plurality of physiological indexes, wherein the target physiological index is any physiological index in the plurality of physiological indexes, adjusting the physiological data of the target physiological index based on the abnormal reason of the target collection device of the physiological data of the target physiological index to obtain adjusted physiological data of the target physiological index, and inputting the adjusted physiological data of the plurality of physiological indexes into a preset deep learning model to obtain health analysis results of the key person output by the preset deep learning model. Further, determining whether the physiological data of the target physiological index is abnormal or not and determining the reason why the target acquisition device is abnormal according to the abnormality, wherein the physiological data characteristic information comprises physiological data statistical characteristics, physiological data frequency domain characteristics and physiological data time domain characteristics of each of a plurality of preset abnormal reasons of the target physiological index, determining first similarity between the physiological data of the target physiological index and the physiological data characteristic information of the preset abnormal reasons according to each of the preset abnormal reasons, and determining whether the physiological data of the target physiological index is abnormal or not and determining the reason why the target acquisition device is abnormal according to the first similarity of the preset abnormal reasons. On the basis, the application further provides that whether the physiological data of the target physiological index is abnormal or not and the abnormal reasons of the abnormal target acquisition device are determined according to the first similarities of the plurality of preset abnormal reasons, wherein the method comprises the steps of determining that the physiological data of the target physiological index is abnormal when the largest first similarity of the first similarities of the plurality of preset abnormal reasons is larger than a preset similarity threshold value; the method comprises the steps of taking a preset abnormal reason with first similarity larger than a preset similarity threshold value as an initial abnormal reason, obtaining a first information set of the initial abnormal reason according to each initial abnormal reason, wherein the first information set comprises a plurality of pieces of first information, the first information comprises physiological data characteristic information of an associated physiological index associated with a target physiological index in the plurality of physiological indexes when the target physiological index is abnormal due to the initial abnormal reason, the physiological data characteristic information comprises physiological data statistical characteristics, physiological data frequency domain characteristics and physiological data time domain characteristics, determining second similarity of physiological data associated with the physiological index and the first information according to each piece of first information in the