JMIR Publications

ABSTRACT

Background:

Diabetic kidney disease (DKD) and diabetic retinopathy (DR) are major diabetic microvascular complications contributing to significant morbidity, disability, and mortality worldwide. The kidney and the eye, having similar microvascular structure, physiological and pathogenic features, may experience similar metabolic changes in diabetes.

Objective:

Using machine learning methods integrated with metabolic data to identify biomarkers associated with DKD and DR in a multi-ethnic Asian population with diabetes, and to improve the performance of DKD/DR detection models beyond traditional risk factors.

Methods:

We used machine learning algorithms (LASSO logistic regression and gradient boosting decision tree) to analyze 2,772 adults with diabetes from Singapore Epidemiology of Eye Diseases study, a population-based cross-sectional study conducted in Singapore (2004-2011). From 220 circulating metabolites and 19 extended risk factors, we selected the most important variables associated with DKD (defined as an estimated glomerular filtration rate < 60 mL/min/1.73m^2) and DR (defined as an Early Treatment Diabetic Retinopathy Study severity level >= 20). DKD/DR detection models were developed based on the variable selection results and externally validated on a sample of 5,843 participants with diabetes from UK Biobank (2007-2010). Machine-learned model performance (AUC with 95% confidence interval, sensitivity, and specificity) was compared to that of traditional logistic regression adjusted for age, gender, diabetes duration, hemoglobin A1c, systolic blood pressure, and body mass index.

Results:

SEED participants had a median age of 61.7 years, with 49.1% being female, 20.2% having DKD, and 25.4% having DR. UK Biobank participants had a median age of 61.0 years, with 39.2% being female, 6.4% having DKD, and 5.7% having DR. Machine learning algorithms identified diabetes duration, insulin usage, age, and tyrosine as the most important factors of both DKD and DR. DKD was additionally associated with cardiovascular disease history, anti-hypertensive medication use, and three metabolites (lactate, citrate, and cholesterol esters to total lipids ratio in intermediate-density-lipoprotein); While DR was additionally associated with hemoglobin A1c, blood glucose, pulse pressure, and alanine. Machine-learned models for DKD and DR detection outperformed the traditional logistic regression model in both internal (AUC: 0.838 vs. 0.743 for DKD, and 0.790 vs. 0.764 for DR) and external validation (AUC: 0.790 vs. 0.692 for DKD, and 0.778 vs. 0.760 for DR).

Conclusions:

Machine learning highlighted duration of diabetes, insulin usage, age, and circulating tyrosine as the most important factors for the detection of diabetic microvascular complications. Machine learning integrated with biomedical big data enables biomarker discovery and improves disease detection beyond traditional risk factors. Clinical Trial: NA