Diabetes predicting model construction method based on electronic medical record data mining

Abstract

The invention discloses a diabetes predicting model construction method based on electronic medical record data mining. The method comprises the steps of performing electronic medical record data cleaning, constructing a predicting model, performing data integration on the electronic medical record data which are introduced from each server through a unique medical case number, wherein the data comprise basic information, diagnosis data, saccharification testing data and biochemical testing data, and integrating the basic information, the diagnosis information and the like for obtaining a whole sample. Data cleaning is performed for eliminating abnormal data, repeated data and existing error data. The cleaned data are stored in a database. Classification prediction is performed on the cleaned diabetes data. Through a result, the classification precision and the model evaluation index of the improved BP neural network model are higher than that of other algorithm models. The method hasadvantages of improving detection rate of undiagnosed diabetic nephropathy in the population, improving diabetic nephropathy preventing effect and saving a large amount of sanitation resource.

Description

technical field [0001] The invention belongs to the field of chronic disease risk prediction of medical big data mining, and relates to a method for building a model, in particular to a method for building a chronic disease risk prediction model based on electronic medical record data mining. Background technique [0002] Diabetes is a relatively common chronic disease that endangers people's healthy life. Diabetes has become the third chronic disease that threatens human health after cardiovascular and cerebrovascular diseases and malignant tumors. With the general improvement of people's living standards and the acceleration of the pace of life in our country, the number of diabetic patients is increasing at an alarming rate, and is developing towards a younger age. The latest survey shows that there are as many as 114 million adult diabetic patients in my country, showing a high incidence rate, but the awareness rate, treatment rate and compliance rate are low. Diabetic...

AI Technical Summary

Problems solved by technology

The number of neural network layers will be selected according to the sample size. Once the sample size data is too small, the recognition error rate will be greatly increased.

Images