Fig. 2: Machine learning-predicted insulin resistance enables improved diabetes stratification compared to previously established metrics.
From: Machine learning-predicted insulin resistance is a risk factor for 12 types of cancer
a Receiver operating characteristic (ROC) curve illustrating the predictive performance of AI-IR, BMI, Metabolic Syndrome, TG/HDL ratio, and TyG index for the incidence of diabetes during the follow-up period. The area under the curve (AUC) for each metrics is also shown. b–e Effect of AI-IR and BMI (b), Metabolic syndrome (MetS) (c), TG/HDL ratio (d), or TyG index (e) on the odds ratio (OR) for onset of diabetes during the follow-up period in participants without diabetes at baseline, adjusted for age and sex. Logistic regression was used with a two-sided Wald test. In the graph, points represent the estimated OR, and error bars represent 95% confidence intervals. Source data are provided as a Source data file.
