Predictive model and risk analysis for diabetic retinopathy using machine learning: a retrospective cohort study in China.

6–8 However, the ordinarily used statistical methods, including logistic regression, show the over-fitting and instability of coefficients when a number of intercorrelated biomarkers are used and thus many practically significant factors are not supported by statistical results due to the limitations. 9 Machine learning algorithms that have better generalisability and discrimination in high-dimensional data can prevent the samples from following the strict inclusion and exclusion criteria, thus reflecting the real health status of all patients.

also in 1,558 other papers