Machine learning model

Machine learning model improves F1 score in silico (absolute increase of 9.13%).

“our model achieved a 57.42% F1 score, overperforming ADA Risk Calculator’s 48.29% by an absolute increase of 9.13%.”

In silicodiabetes screening without lab testsn = 38,476

Leveraging Routinely Collected Health Data and Social and Behavioral Determinants of Health in Building Machine-Learning Tools for Diabetes Screening

ADA Scientific Sessions 2025 abstract·13 Jun 2025

Machine learning model improves sensitivity in silico (sensitivity 71.56% vs ADA Risk Calculator 48.85%).

“Leveraging LUPI, our model further improved sensitivity to 71.56%.”

In silicodiabetes screening without lab testsn = 38,476

Leveraging Routinely Collected Health Data and Social and Behavioral Determinants of Health in Building Machine-Learning Tools for Diabetes Screening

ADA Scientific Sessions 2025 abstract·13 Jun 2025