Unique predictors in fitted Gene&Binary&Count elastic net model by antibiotic.

The important predictors included in the fitted Gene&Binary&Count models for each antibiotic were identified and compared to the shared important predictors for antibiotics within the same class. For each antibiotic class, “Shared” is the number of important predictors selected in all antibiotics within the same class, the denominator indicates the total number shared predictors. is the total number of important predictors selected by the model. is the sum of coefficient absolute values for predictors. refers to the number of candidate predictors considered. refers to the number of candidate predictors with estimated power ≥0.80 at an effect size of 2.3. “Unique” is the number of important predictors that are unique to that antibiotic relative to its class. β-lactams and β-lactam combination antibiotics had the highest number of total and unique important predictors spread across each of the antibiotics in its class. Models for fluoroquinolone resistance only agreed on a single predictor (Count).