Diabetes early warning method, system and device based on self-supervised DNN and storage medium

Abstract

The invention discloses a diabetes early warning method based on self-supervised DNN. The method the following steps: performing data preprocessing on original body index data, including label processing, numeralization processing, standardization processing, missing value processing and feature value selection processing; designing three deep neural networks, and carrying out model batch iteration training by taking the processed data as input; carrying out the data processing steps on the original body index data of a user to be predicted, inputting the data into the trained model, and carrying out diabetes disease risk visualization early warning. The invention further discloses a diabetes early warning system based on the self-supervised DNN, computer equipment and a computer readable storage medium. According to the self-supervised learning method, deep association of user health indexes can be mined through a small amount of label data, the designed model can adapt to other prediction tasks, and the method has the advantages of reliability, universality and the like.

Description

technical field [0001] The present invention relates to the fields of artificial intelligence, big data medical treatment, etc., and in particular to a self-supervised DNN-based diabetes early warning method, system, computer equipment and computer-readable storage medium. Background technique [0002] In recent years, the number of patients with chronic diseases in my country ranks first in the world, such as diabetes, cardiovascular disease and other chronic diseases. As a common chronic disease, diabetes and its related complications are an important part of it. At present, there is no cure, but scientific and effective intervention, prevention and treatment can be used to reduce the incidence rate and improve the quality of life of patients. In the field of diabetes risk early warning systems, traditional machine learning methods are often based on simple tree models for prediction, lack of modeling of complex associations of user health data, and rely heavily on a large...

AI Technical Summary

Problems solved by technology

The main problems to be solved by the present invention are: first, how to use a small amount of labeled data for prediction; second, how to deeply correlate prediction results with user health indicators; third, how to flexibly adjust for different data characteristics

Images