Table 2 Effect of workload and effect of predicted fatigue on yield of testing for heart attack (more precisely, the positive rate of tests done in the few days after the ED visit) (n = 1017)

From: A machine learning model using clinical notes to identify physician fatigue

 

Effect of workload

Effect of predicted fatigue

 

on yield of testing

on yield of testing

Coefficient

−0.0081

−0.2935*

 

(0.009)

(0.147)

Controls

  

 Time of day

YES

YES

 Day of week

YES

YES

 Week of year

YES

YES

 Year

YES

YES

 Demographics

YES

YES

 Chief complaint

YES

YES

 Physician

YES

YES

 Length of stay

YES

YES

 Insurance type

YES

YES

  1. Column 1 shows a regression of testing yield on a physician’s prior week workload, which is not significant (p = 0.383 in a two-sided test). Column 2 shows a regression of testing yield on the predicted fatigue score, calculated on the basis of the patient’s note, which is significant and negative (p = 0.047 in a two-sided test).
  2. p < 0.05*, p < 0.01**, p < 0.001***.
Back to article page