Abnormal fasting blood glucose level early-warning method based on integrated learning fusion model

Abstract

The invention discloses an abnormal fasting blood glucose level early-warning method based on an integrated learning fusion model. With combination of individual physical examination data of blood routine examination, liver function, blood fat, and renal function and the like, a fasting blood glucose level is predicted by fusing a gradient regression tree, a random forest, and a linear regressionmodel and the like based on an integrated learning method. The prediction model is trained by using lots of training data, so that the accuracy, universality and robustness of the prediction model areimproved. Fasting blood glucose level prediction is carried out on timely an individual without fasting blood glucose checking, so that the high-diabetes-risk patient is warned early and effectively.

Description

technical field [0001] The invention relates to the fields of smart medical care and machine learning, and in particular to an early warning method for abnormal fasting blood sugar levels based on an integrated learning fusion model. Background technique [0002] With the rapid economic development and the acceleration of industrialization, the change of lifestyle and the acceleration of the aging process, the prevalence of diabetes in my country is showing a rapid upward trend, which is called another serious hazard after cardiovascular and cerebrovascular diseases and tumors. Important chronic non-communicable diseases of people's health. According to the Chinese Type 2 Diabetes Prevention Guidelines (2013 Edition), it is estimated that from 2005 to 2015, the economic loss caused by diabetes and related cardiovascular diseases in China reached 557.7 billion US dollars. damage and lead to shortened life expectancy, but also bring a heavy economic burden to individuals and c...

AI Technical Summary

Problems solved by technology

[0006] However, in terms of the physiological mechanism of the human body, fasting blood glucose has a complex relationship with other physiological indicators, and the physical examination indicators (features) used to train the random forest model are insufficient, and there is a risk of large deviations in the prediction of fasting blood glucose values

Secondly, the random forest model used as a single prediction model has a large deviation in the prediction of continuous values, and the prediction accuracy needs to be further improved

In addition, this technology judges the risk of diabetes on the basis of the difference between the fasting blood glucose value of the next year and the current fasting blood glucose value, and does not take into account the quantitative relationship between the specific value of fasting blood glucose and diabetes

Images