Artificial Intelligence in the Initial Assessment of Acute Burn Wounds: : A Narrative Literature Review

Andre Parmonangan (Universitas Lampung)
Lucretya Yeniwati Tanuwijaya (Reconstructive, and Aesthetic Surgery, St. Elisabeth Hospital Semarang)

Jurnal Kesehatan dan Kedokteran
Vol. 5 No. 2 (2026): Juni: Jurnal Kesehatan dan Kedokteran

Accurate initial assessment of acute burn wounds is essential for determining severity, healing potential, referral needs, and treatment strategy. However, this assessment remains challenging because burn wounds evolve over time and conventional methods often depend on clinician experience or specialized imaging technologies. Artificial intelligence offers a potential approach to improve the objectivity and efficiency of early burn assessment. Objective: This study aims to analyze current applications of artificial intelligence in the initial assessment of acute burn wounds and evaluate its potential, methodological challenges, and clinical implementation. Methodology: This study employed a narrative literature review. Scientific publications were retrieved from PubMed and Google Scholar from database inception to May 2026. Eligible publications included original research articles, systematic and narrative reviews, clinical practice guidelines, and AI reporting guidelines related to AI-assisted burn assessment. The selected literature was synthesized narratively by comparing evidence on burn detection, burn-depth classification, wound segmentation, total body surface area estimation, prediction of surgical requirements, multimodal AI, and methodological quality. Findings: AI has demonstrated promising performance in burn detection, burn-depth classification, wound segmentation, total body surface area estimation, and prediction of surgical requirements. Recent studies have also expanded AI applications toward multimodal assessment by integrating clinical images with physiological and electronic medical record data. Nevertheless, current evidence is constrained by heterogeneous reference standards, retrospective designs, limited external validation, inconsistent image-acquisition protocols, and inadequate representation of diverse skin tones. Implications: AI should be implemented as a clinical decision-support tool that complements, rather than replaces, serial clinical assessment by burn specialists. Originality: This review integrates technical advances in artificial intelligence with clinically relevant considerations, including burn wound progression, image-acquisition timing, treatment status, reference-standard heterogeneity, external validation, skin-tone diversity, and workflow integration.