Skip to main content

Table 3 Features selected in the XGboost model

From: Predicting 30-days mortality for MIMIC-III patients with sepsis-3: a machine learning approach using XGboost

  OR_with_CI P
(Intercept) 493.907 (9.063–27,931.087) 0.00247
Urineoutput 1 (1–1) < 0.001
Lactate_min 1.401 (1.288–1.527) < 0.001
Bun_mean 1.018 (1.013–1.023) < 0.001
Sysbp_min 0.979 (0.974–0.984) < 0.001
Metastatic_cancer 2.997 (2.217–4.038) < 0.001
Inr_max 1.058 (1.002–1.115) 0.03709
Age 1.019 (1.013–1.025) < 0.001
Sodium_max 1.016 (1.001–1.031) 0.03835
Aniongap_max 1.048 (1.026–1.069) < 0.001
Creatinine_min 0.766 (0.686–0.852) < 0.001
Spo2_mean 0.897 (0.865–0.93) < 0.001
  1. OR odds ratio, CI confidence interval, bun blood urea nitrogen, sysbp systolic blood pressure, INR international normalized ratio, Spo2 oxyhemoglobin saturation, Max maximum, Min minimum