Table 1.

Selection criteria of predictive modeling studies in PICOTS format.

	Participants (P)	Intervention (I)	Comparison (C)	Outcomes (O)	Timeframe (T)	Setting (S)	Other limits
Inclusion criteria	Individuals with T2DM Individuals without T2DM being investigated for the condition	ML predictive modeling: supervised, unsupervised, semi-supervised ML, or combinations thereof	Not applicable	Primary: metrics of discrimination ability, calibration, and classification accuracy in T2DM prediction Secondary: candidate predictors, applied algorithms, level of validation, intended use of models	Since 1 January 2009 to date	Clinical care settings, for example, hospitals, long-term-, ambulatory-, acute-care facilities Community care settings, e.g. general practices, community health centers, allied health practices	Language  =  English
Exclusion criteria	Patients with other clinical phenotypes of diabetes, type 1 diabetes, gestational diabetes Pre-diabetic individuals Diabetic complications	Predictive modeling without an explicit ML approach				Laboratory settings: using only genetic, genomic, or genotype data

ML: machine learning; T2DM: type 2 diabetes.