Skip to main content
. 2022 Jan 12;12(1):e055918. doi: 10.1136/bmjopen-2021-055918

Table 3.

Test characteristics of prediction models for outcomes on test cohort

Outcome Model Specificity (95% CI) Sensitivity (95% CI) PPV (95% CI) NPV (95% CI) Cut-off
TBI
LR 47.5 (40.9 to 54.0) 80.3 (68.7 to 89.1) 29.9 (23.3 to 37.3) 89.6 (82.9 to 94.3) 0.136
XGB 72.5 (66.3 to 78.1) 80.3 (68.7 to 89.1) 44.9 (35.7 to 54.3) 92.9 (88.2 to 96.2) 0.268
SVM 64.8 (58.4 to 70.9) 80.3 (68.7 to 89.1) 39.0 (30.7 to 47.7) 92.2 (87.0 to 95.8) 0.191
RF 68.2 (61.9 to 74.1) 80.3 (68.7 to 89.1) 41.4 (32.8 to 50.4) 92.5 (87.6 to 96.0) 0.185
EN 61.0 (54.5 to 67.3) 80.3 (68.7 to 89.1) 36.6 (28.7 to 44.9) 91.7 (86.3 to 95.5) 0.205
TBI-I
LR 71.1 (65.0 to 76.7) 80.4 (67.6 to 89.8) 38.8 (29.9 to 48.3) 94.1 (89.7 to 97.0) 0.164
XGB 74.0 (68.0 to 79.4) 80.4 (67.6 to 89.8) 41.3 (31.9 to 51.1) 94.3 (90.0 to 97.1) 0.143
SVM 71.1 (65.0 to 76.7) 80.4 (67.6 to 89.8) 38.8 (29.9 to 48.3) 94.1 (89.7 to 97.0) 0.172
RF 76.0 (70.2 to 81.2) 80.4 (67.6 to 89.8) 43.3 (33.6 to 53.3) 94.4 (90.3 to 97.2) 0.205
EN 81.3 (75.9 to 86.0) 80.4 (67.6 to 89.8) 49.5 (38.8 to 60.1) 94.8 (90.9 to 97.4) 0.204
TBI-ND
LR 46.1 (39.8 to 52.6) 80.7 (68.1 to 90.0) 25.8 (19.6 to 32.9) 91.1 (84.7 to 95.5) 0.090
XGB 66.5 (60.2 to 72.4) 80.7 (68.1 to 90.0) 35.9 (27.7 to 44.9) 93.7 (89.0 to 96.8) 0.242
SVM 59.2 (52.7 to 65.4) 80.7 (68.1 to 90.0) 31.5 (24.1 to 39.7) 92.9 (87.7 to 96.4) 0.147
RF 60.4 (54.0 to 66.6) 80.7 (68.1 to 90.0) 32.2 (24.6 to 40.5) 93.1 (88.0 to 96.5) 0.138
EN 74.3 (68.3 to 79.6) 80.7 (68.1 to 90.0) 42.2 (32.8 to 52.0) 94.3 (90.0 to 97.1) 0.201
TBI-D
LR 42.6 (36.9 to 48.5) 81.8 (48.2 to 97.7) 5.1 (2.4 to 9.5) 98.4 (94.4 to 99.8) 0.005
XGB 57.7 (51.8 to 63.5) 81.8 (48.2 to 97.7) 6.8 (3.2 to 12.5) 98.8 (95.8 to 99.9) 0.002
SVM 74.2 (68.8 to 79.2) 81.8 (48.2 to 97.7) 10.7 (5.0 to 19.4) 99.1 (96.7 to 99.9) 0.039
RF 74.9 (69.5 to 79.8) 81.8 (48.2 to 97.7) 11.0 (5.1 to 19.8) 99.1 (96.8 to 99.9) 0.005
EN 79.0 (73.9 to 83.6) 81.8 (48.2 to 97.7) 12.9 (6.1 to 23.0) 99.1 (96.9 to 99.9) 0.033

EN, elastic net; LR, logistic regression analysis; RF, random forest; SVM, support vector machine; TBI, traumatic brain injury; TBI-D, traumatic brain injury with death; TBI-I, traumatic brain injury with intracranial injury; TBI-ND, traumatic brain injury with non-discharge; XGB, extreme gradient boosting.