Advancing polytrauma care: developing and validating machine learning models for early mortality prediction

Table 1 Baseline characteristics of training, internal, and external validation sets

	Training set (N = 2153)	Internal validation set (N = 922)	External validation set (N = 1673)	P Value^c
Demographic characteristics
Age^a, years	43 (28, 61)	43 (29, 60)	42 (28, 58)	0.106
Sex^b				0.482
Female	555 (25.8)	257 (27.9)	441 (26.4)
Male	1598 (74.2)	665 (72.1)	1232 (73.6)
BMI^a, kg/m^2	24.6 (23.3, 26.0)	24.6 (23.4, 26.0)	24.5 (22.4, 26.3)	0.339
Scoring system
ISS^a	29 (22, 38)	27 (22, 38)	27 (22, 35)	0.076
GCS^a	7 (3, 15)	7 (3, 14)	13 (3, 15)	< 0.001
Laboratory test
pH^a	7.34 (7.28, 7.37)	7.34 (7.28, 7.38)	7.34 (7.29, 7.38)	0.020
BE^a, mmol/L	− 2.80 (− 5.40, − 1.15)	− 2.98 (− 5.40, − 1.15)	− 2.67 (− 4.90, − 1.00)	0.028
Lactate^a, mmol/L	2.30 (1.50, 3.40)	2.30 (1.50, 3.54)	2.20 (1.50, 3.20)	0.042

Bold values indicate a p-value of less than 0.05
BMI body mass index, ISS injury severity score, GCS glasgow coma scale, BE base excess
^aValues are presented as the median (inter-quartile range)
^bValues are presented as number (percentage)
^cP values between groups were assessed by the Chi-square and Kruskal–Wallis H tests

ISSN: 1479-5876