To assess the value of quantitative EEG (qEEG) as a diagnostic and prognostic biomarker in infants with abusive head trauma (AHT). Despite its central role in monitoring encephalopathy, EEG remains underused in multimodal evaluations, and its quantitative analysis may provide objective, real-time insights into cerebral dysfunction and long-term outcome.

This retrospective monocentric case–control study included infants under two years with confirmed AHT and age- and sex-matched controls. Clinical and early EEG data were collected. Patients’ outcome was stratified by Pediatric Overall Performance Category score (POPC _1–3 vs. _4–6 ). Quantitative EEG features were analyzed, and two neural networks were trained using five-fold cross-validation for diagnosis and outcome prediction.

84 EEGs from 75 participants were analyzed (46 EEGs from 40 AHT; 38 EEGs from 35 controls). Compared with controls, AHT EEGs showed significantly reduced entropy and Hurst exponent values and increased low-frequency power, reflecting diffuse cortical dysfunction. Within the AHT group, reduced signal complexity and loss of interhemispheric asymmetry correlated with unfavorable outcomes (POPC _4–6 , p < 0.01). Machine learning perfectly classified AHT cases versus controls and classified patients into POPC _1-3 or POPC _4-6 groups with 73±14 % accuracy. Combined models distinguished control, POPC _1-3 , and POPC _4-6 groups with 90±5 % accuracy.

Early qEEG provides functional information that complements imaging and clinical findings. qEEG-derived biomarkers may enable early risk stratification, guide neuroprotective strategies, and improve prognostic counseling in infants with AHT. Larger multicenter prospective studies are warranted to validate these exploratory findings and define their clinical applicability.