Fig. 3 | Scientific Reports

Fig. 3

From: Blood-based biomarkers suggest prolonged axonal Injury following pediatric mild traumatic brain injury

Fig. 3

Group classification using Random forest machine learning algorithm. Receiver operating characteristics and variable importance (VI) results from the Random forest supervised machine learning algorithm for classifying diagnostic (Dx) status in patients with pediatric mild traumatic brain injury (pmTBI) relative to healthy controls (HC). The model included Post-concussion Symptom Inventory percent total score [PCSI], glial fibrillary acidic protein (GFAP), neurofilament light chain (NFL), ubiquitin C-terminal hydrolase L1 (UCH-L1) and phosphorylated tau 181 (pTau181). Results are presented separately for the first (V1; Panel a) and second (V2; Panel b) visits and include balanced accuracy (BA), specificity (Spe), and sensitivity (Sen). The intersection of optimal sensitivity/specificity is denoted with a circle. NFL contributed more towards diagnostic accuracy relative to PCSI scores at V2.

Back to article page