Skip to main content

Table 3 Median and 2.5 and 97.5 percentiles of difference in under- and overtriage rates between learners (calculation on 1000 runs)

From: The advanced machine learner XGBoost did not reduce prehospital trauma mistriage compared with logistic regression: a simulation study

Events per free parameter Data set Difference in undertriage LogReg-XGBoost Difference in overtriage LogReg-XGBoost
10 SweTrau 0 [− 0.354, 0.025] 0 [− 0.005, 0.265]
10 NTDB − 0.005 [− 0.398, 0.035] 0.004 [− 0.017, 0.433]
25 SweTrau 0 [− 0.301, 0.015] 0 [− 0.003, 0.238]
25 NTDB  − 0.005 [− 0.39, 0.023] 0.003 [− 0.015, 0.427]
100 SweTrau 0 [− 0.025, 0.005] 0 [− 0.001, 0.008]
100 NTDB 0 [− 0.396, 0.005] 0 [− 0.003, 0.427]
1000 NTDB 0 [− 0.386, 0] 0 [− 0.001, 0.422]
  1. NTDB, National Trauma Data Bank; SweTrau, Swedish Trauma Registry