Method for constructing diabetes disease risk model based on scoring system

Abstract

The invention discloses a method for constructing a diabetes disease risk model based on a scoring system, and belongs to the field of big data and machine learning. The method comprises the steps: collecting physical examination data of a target region, preprocessing the data, extracting features from the preprocessed data to obtain a feature set, carrying out feature screening on the obtained feature set to obtain a screened optimal feature subset, binning each feature in the obtained optimal feature subset, calculating the WOE of each box, mapping the WOE into data, modeling by using logistic regression, and making the diabetes mellitus disease risk score card by using a scoring module. The risk model constructed by adopting the method can obtain a finer and quantified diabetes disease risk score, and by applying the model, an individual can be guided to carry out more appropriate individual health management, including more reasonable dietary habits, exercise habits and other living habits, and a more accurate and objective commercial underwriting decision can be made by a commercial insurance mechanism.

Description

technical field [0001] The invention relates to the fields of big data and machine learning, in particular to a method for constructing a scoring system-based diabetes risk model. Background technique [0002] At present, various types of chronic diseases, including various cardiovascular and cerebrovascular diseases, tumors, chronic respiratory diseases, diabetes, etc., have caused serious burdens to society and families, and are showing a rapid increase and a younger trend. [0003] In 2015, the National Health and Family Planning Commission released the "Report on the Progress of Disease Prevention and Control in China (2015)" at the regular press conference. Chronic diseases such as chronic diseases and malignant tumors have become the main cause of death. The number of deaths caused by chronic diseases has accounted for 86.6% of the total deaths in the country, and the disease burden caused by them accounts for nearly 70% of the total disease burden. [0004] The curre...

AI Technical Summary

Problems solved by technology

[0003] In 2015, the National Health and Family Planning Commission released the "Report on the Progress of Disease Prevention and Control in China (2015)" at the regular press conference. Chronic diseases such as chronic diseases and malignant tumors have become the main cause of death. The number of deaths caused by chronic diseases has accounted for 86.6% of the total deaths in the country, and the disease burden caused by them accounts for nearly 70% of the total disease burden.

[0005] At present, the public's awareness of physical examination is gradually increasing. Although the individual physical examination report can provide individuals with a certain degree of health information including the risk of diabetes, the current medical examination report itself and medical staff can give the risk of diabetes from the results of the physical examination report. , based on only one or several clinical laboratory indicators, such as urine sugar, blood sugar, cholesterol, etc., as well as indicators such as body weight, waist circumference, and blood pressure. The risk of diabetes based on these indicators has the following defects: First, , which depends heavily on the experience of medical staff, is highly subjective, and has little reference value; second, the influencing factors of diabetes are very complex, including not only age, gender, genetics, hypertension, diabetes, dyslipidemia, overweight and obesity, Unhealthy diet, lack of physical exercise, smoking, excessive mental stress, excessive alcohol consumption, and social factors such as geography and population, the risk results of diabetes that can be given at present, even if some models are used to evaluate, can only get "yes" ” or “No”, the rough qualitative results are not very meaningful

[0006] In other words, the current risk of diabetes obtained from the physical examination report is rough, with large errors and little reference value, which leads to:

[0007] (1) For individuals, even if they insist on regular physical examinations, they cannot obtain more accurate diabetes risk information, which is not conducive to individual health management;

[0008] (2) For medical institutions such as hospitals and medical examination institutions, since they cannot give more accurate and reliable health management suggestions from individual medical examination reports, the trust of customers gradually decreases;

[0009] (3) For insurance institutions that provide various commercial insurances, the commercial decision of “yes” or “no” can only be drawn from the medical examination report as to whether to underwrite the insurance and the pricing of premiums. However, such decisions are based on ideals. Information is unreliable, and premiums cannot be priced according to different risk levels, causing unfair problems to both customers and insurance institutions themselves, and commercial and social benefits need to be improved;

[0010] (4) Local governments at all levels and national health care departments cannot accurately grasp the diabetes risk of people in their jurisdictions, so that they cannot formulate more reasonable social medical insurance policies and overall management of social health

Images