BURNS-PBIA (Burns Platform Based on Artificial Intelligence): A smartphone-based system for early burn depth assessment, TBSA estimation, and surgical decision prediction

Vaca B.; Mayer H.F.; Ciudad P. · European Journal of Plastic Surgery (2026) · DOI: 10.1007/s00238-026-02415-9

Resumen

Background: Clinical assessment of partial-thickness burns is limited by variability among observers. Objective tools are needed to discriminate superficial (IIA) from deep partial-thickness (IIB) injuries and support early management. This study evaluates the diagnostic performance and clinical utility of BURNS-PBIA, Burns Platform Based on Artificial Intelligence, a Vision Transformer–based model for burn-depth classification, TBSA estimation, and therapeutic prediction using standardized smartphone images. Methods: Retrospective analysis of 320 images from 80 adults (January 2024–January 2025). Photographs were obtained within 24 hours following a uniform imaging protocol. Burn depth was determined by three certified surgeons (κ = 0.86). A ViT-B/16 model pretrained on ImageNet was fine-tuned with a patient-level 70/15/15 split. Evaluation metrics included AUC, sensitivity, specificity, F1-score, MAE for TBSA, calibration analysis, and decision-curve analysis. Results: The proposed model yielded an AUC of 0.93 (95% CI 0.88–0.98), sensitivity 90%, specificity 85%, F1-score 0.87, and Brier score 0.07. TBSA estimation achieved a low MAE. Predictions of grafting and surgical intervention showed high concordance with observed outcomes. Conclusion: BURNS-PBIA is a clinically oriented decision-support system that provides accurate and well-calibrated predictions for burn depth classification, TBSA estimation, and early management planning using standardized smartphone photographs. The platform demonstrates strong diagnostic performance and practical applicability within routine burn care workflows. External multicenter validation is warranted prior to widespread clinical deployment. Level of evidence: Level III, diagnostic study. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2026.

Instituciones: Universidad Nacional Mayor de San Marcos