이진석 (경희대학교) | 한빛사논문 > 한빛사

한빛사논문

조회64

경희대학교

CV File

EBioMedicine, Mar 13 2026, 126 106206 S2352-3964(26)00088-5 | https://doi.org/10.1016/j.ebiom.2026.106206 Accurate, fair, and generalisable scaling of injury severity score-based AI with demographics in terms of mortality in patients with trauma: multi-centre, multi-national retrospective cohort study

Yunjeong Choi^a,n, Junepill Seok ^b,c,n, Thomas Young-Chul Oh ^d, Jeremy Hsu^d, Do Wan Kim^e, Byungchul Yu ^f,g, Jayun Cho ^g, Woocheol Jang^a, Jina Kim ^a, Na-eun Oh ^a, Jehyeuk Ahn ^a, Robert J Femia^h, Paul A Testa^h, Dong Keon Yon^i,j, Daniel K Sodickson ^k,l, Wu Seong Kang ^f,g,m,∗, Jinseok Lee ^a,∗

^aDepartment of Biomedical Engineering, Kyung Hee University, Yongin, Republic of Korea
^bDepartment of Thoracic and Cardiovascular Surgery, Chungbuk National University Hospital, Cheongju, Republic of Korea
^cDepartment of Thoracic and Cardiovascular Surgery, Chungbuk National University, College of Medicine, Cheongju, Republic of Korea
^dTrauma Service, Westmead Hospital, University of Sydney, NSW, Australia
^eDepartment of Thoracic and Cardiovascular Surgery, Chonnam National University Hospital and Chonnam National University Medical School, Gwangju, Republic of Korea
^fDepartment of Traumatology, Gachon University College of Medicine, Incheon, Republic of Korea
^gDepartment of Trauma Surgery, Gachon University Gil Medical Center, Incheon, Republic of Korea
^hRonald O. Perelman Department of Emergency Medicine, New York University Grossman School of Medicine, New York, NY, USA
ⁱCentre for Digital Health, Medical Science Research Institute, Kyung Hee University College of Medicine, Seoul, South Korea
^jDepartment of Precision Medicine, Kyung Hee University College of Medicine, Seoul, South Korea
^kDepartment of Radiology, Bernard and Irene Schwartz Centre for Biomedical Imaging, New York University Grossman School of Medicine, New York, NY, USA
^lCentre for Advanced Imaging Innovation and Research (CAI2R), Department of Radiology, New York University Grossman School of Medicine, New York, NY, USA
^mDepartment of Trauma Surgery, Jeju Regional Trauma Centre, Cheju Halla General Hospital, Jeju, Republic of Korea
ⁿThese authors contributed equally to this work and should be considered co-first authors.

^*Corresponding authors: correspondence to Wu Seong Kang or Jinseok Lee

Abstract

Background
Accurate and equitable prediction of trauma-related in-hospital mortality is critical for guiding clinical decisions and optimising trauma care resources. Traditional severity scoring systems like the Injury Severity Score (ISS) do not account for demographic factors, potentially limiting their fairness and generalisability across diverse populations.

Methods
We developed and externally validated an artificial intelligence (AI) model based on ISS and integrated demographic features (age and sex) to predict in-hospital mortality after trauma. Data from the Korean Trauma Data Bank were used for model development and internal validation, comprising 121,418 patients with trauma aged ≥15 years treated at 19 trauma centres in South Korea (2017–2022). External validation was performed on an independent cohort of 7458 patients from five trauma centres (four in South Korea and one in Australia, 2022–2024). The primary outcome was trauma-related in-hospital mortality. Predictive performance was assessed using area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, accuracy, and balanced accuracy. Fairness was evaluated by comparing AUROC differences across age (<65 vs ≥65 years) and sex (female vs male) subgroups.

Findings
The ISS-based AI model incorporating age and sex achieved high predictive performance (internal validation AUROC, 0.934; external validation AUROC range, 0.901–0.920), outperforming conventional ISS-based methods. The model also demonstrated improved fairness, showing reduced AUROC differences across subgroups (age: 0.068 vs 0.091; sex: 0.021 vs 0.046 for AI model vs ISS, respectively).

Interpretation
Scaling an ISS-based AI model through demographic integration yielded accurate, fair, and generalisable predictions of trauma-related in-hospital mortality. This approach may enhance trauma care decision-making and enable more equitable resource allocation across diverse clinical settings.

논문정보