Table 1

Measurements used as inputs for machine learning algorithms (MLAs) and for calculating CDC risk score.

Measurements used as inputs to MLAMeasurements used as inputs to CDC risk score
Clinical measurementsBMINone
Blood pressure (systolic and diastolic)
Laboratory valuesBlood urea nitrogenNone
Cholesterol (HDL and LDL)
White cell count
Medical historyPresence of past acute kidney injuryCardiovascular disease
History of chronic heart failureCongestive heart failure
Reported smoking historyPeripheral vascular disease
Reported alcohol historyProteinuria
  • Demographics (age, sex), clinical measurements (BMI, blood pressure (systolic and diastolic)), laboratory values (blood urea nitrogen, creatinine and eGFR, cholesterol (high-density lipoprotein and low-density lipoprotein), white cell count), and medical history (presence of past acute kidney injury, history of chronic heart failure, reported smoking history, reported alcohol history) served as input features for the MLA models. The clinical and laboratory measurement values were pooled using 5th and 95th percentiles, median, and last available result over 1 year prior to T2DM diagnosis.

  • BMI, body mass index; CDC, Centers for Disease Control and Prevention; eGFR, estimated glomerular filtration rate; HDL, high-density lipoprotein; LDL, low-density lipoprotein; T2DM, type 2 diabetes mellitus.