resumen

Atef Eid Madkour Elsayeda*, Mohsin Shahabuddin, Abdulrahman Emad Shafie, Fadi Elshafi Babikir Mohammed*, Abid Waseem, Muneer Ahmed Mohammed Bakr, Waheed Ibrahim Mohamed Alasiri, Ghaliah Waleed Hamid Othman, Sayyaf Mohammed Alhazmi, Imtiaz Ahmed S/0 Muhammad Soomar Soomro

Background: Artificial intelligence (AI) has emerged as a transformative tool in emergency medicine, where timely and accurate decision-making is vital. Machine learning (ML), deep learning (DL), and natural language processing (NLP) approaches are increasingly applied to enhance triage, risk stratification, and prediction of adverse outcomes. Despite these advances, variability in methodologies and reported outcomes underscores the need for a systematic synthesis of the evidence.

Objective: This systematic review evaluates the impact of AI in emergency medicine, focusing on its influence on triage accuracy, clinical outcomes, workflow efficiency, and overall patient care.

Methods: The review was conducted in line with PRISMA 2020 guidelines. Eligible studies included randomized controlled trials, retrospective and prospective cohorts, and case-control designs assessing AI use in emergency or critical care. Literature searches were performed across PubMed, Scopus, Web of Science, Embase, and IEEE Xplore (2010–2024). Two independent reviewers conducted screening, data extraction, and risk of bias assessment. Due to heterogeneity, findings were synthesized narratively.

Results: Fifteen studies met inclusion criteria. AI models achieved high predictive accuracy in mortality (AUROC >0.92 in traumatic brain injury), sepsis identification, hospital admission, cardiac arrest, and triage in both adults and children. NLP-based systems improved acuity assignment by leveraging unstructured clinical narratives. Consistently, AI outperformed conventional triage tools in sensitivity, efficiency, and early risk recognition. Nonetheless, most studies were retrospective, single-centre, and lacked external validation, limiting generalizability.

Conclusions: AI demonstrates significant potential to advance emergency medicine by strengthening triage and prediction capabilities. To ensure safe translation into practice, future efforts must emphasize prospective multicentre trials, explainable models, and equitable integration into clinical workflows.