Evaluation of Artificial Intelligence (ChatGPT-5.2) in the Classification and Indication for Fixation of Posterior Malleolar Fractures: A Multicenter External Validation Study
DOI:
https://doi.org/10.15417/issn.1852-7434.2026.91.3.2348Keywords:
Artificial intelligence, posterior malleolus, multicenter studyAbstract
Introduction: Posterior malleolar fractures have a significant impact on ankle joint congruity. The indication for fixation no longer depends solely on fragment size but also on fracture morphology. Artificial intelligence (AI) has emerged as a tool to support clinical decision-making. The objective of this study was to evaluate the ability of AI to classify posterior malleolar fractures and determine the indication for fixation, compared with a reference standard based on expert consensus.
Materials and Methods: A retrospective diagnostic accuracy study with external validation was conducted in accordance with the STARD-AI and GAMER guidelines. A protocol based on the Bartoníček and Rammelt classification was developed using 24 cases for calibration. Subsequently, 9 cases were evaluated using radiographs and computed tomography scans and analyzed by 12 experts and the ChatGPT-5.2 model. Agreement in fracture classification and sensitivity for the indication for fixation were assessed using Cohen’s kappa coefficient.
Results: ChatGPT-5.2 achieved 78% agreement in fracture classification, with a kappa coefficient of 0.56, indicating moderate agreement. Sensitivity for the indication for posterior malleolar fixation was 100%.
Conclusions: Artificial intelligence demonstrated performance comparable to that of experts in the classification of posterior malleolar fractures and high sensitivity in determining the indication for fixation. It proved useful as a supportive tool in medical education settings. Studies with larger sample sizes are needed to validate these findings.
Downloads
References
Terstegen J, Weel H, Frosch KH, Rolvien T, Schlickewei C, Mueller E. Classifications of posterior malleolar
fractures: a systematic literature review. Arch Orthop Trauma Surg 2023;143(7):4181-220. https://doi.org/10.1007/s00402-022-04643-7
Mohamed A, Fuad U, Elasad A, Shrestha S, Hagroo A, Pengas IP. Posterior malleolar fractures: From the „Forgotten Fragment“ to modern concepts in management. Cureus 2025;17(10):e94681. https://doi.org/10.7759/cureus.94681
Bartoníček J, Rammelt S, Tuček M, Naňka O. Posterior malleolar fractures of the ankle. Eur J Trauma Emerg Surg 2015;41(6):587-600. https://doi.org/10.1007/s00068-015-0560-6
Verhage SM, Hoogendoorn JM, Krijnen P. When and how to operate the posterior malleolus fragment in trimalleolar fractures. Arch Orthop Trauma Surg 2018;138(9):1213-22. https://doi.org/10.1007/s00402-018-2949-2
Gale W, Oakden-Rayner L, Carneiro G, Bradley AP, Palmer LJ. Detecting hip fractures with radiologist-level performance using deep neural networks. Preprint. Digit Med 2017. https://doi.org/10.48550/arXiv.1711.06504
Lindsey R, Daluiski A, Chopra S, Lachapelle A, Mozer M, Sicular S, et al. Deep neural network improves fracture detection by clinicians. Proc Natl Acad Sci USA 2018;115(45):11591-6. https://doi.org/10.1073/pnas.1806905115
Rivadeneira Jurado HA, Rivadeneira Jurado EA, Espinoza Freire D, Samaniego AF, Lulkin E, Bidolegui F, et al. Evaluación de la clasificación de las fracturas de platillo tibial según Schatzker-Kfuri utilizando radiografías y tomografía. Comparación entre el observador experto y el modelo ChatGPT-4o. Rev Asoc Argent Ortop Traumatol 2025;90(6):556-60. https://doi.org/10.15417/issn.1852-7434.2025.90.6.2224
Husarek J, Hess S, Razaeian S, Ruder TD, Sehmisch S, Müller M, et al. Artificial intelligence in commercial fracture detection products: a systematic review and meta-analysis of diagnostic test accuracy. Sci Rep 2024;14(1):23053. https://doi.org/10.1038/s41598-024-73058-8
Mohammadi S, Parviz S, Parvaz P, Pirmoradi MM, Afzalimoghaddam M, Mirfazaelian H. Diagnostic performance of ChatGPT in tibial plateau fracture in knee X-ray. Emerg Radiol 2025;32(1):59-64. https://doi.org/10.1007/s10140-024-02298-y.
Downloads
Published
License
Manuscript acceptance by the Journal implies the simultaneous non-submission to any other journal or publishing house. The RAAOT is under the Licencia Creative Commnos Atribución-NoComercial-Compartir Obras Derivadas Igual 4.0 Internacional (CC-BY-NC.SA 4.0) (http://creativecommons.org/licences/by-nc-sa/4.0/deed.es). Articles can be shared, copied, distributed, modified, altered, transformed into a derivative work, executed and publicly communicated, provided a) the authors and the original publication (Journal, Publisher and URL) are mentioned, b) they are not used for commercial purposes, c) the same terms of the license are maintained.
In the event that the manuscript is approved for its next publication, the authors retain the copyright and will assign to the journal the rights of publication, edition, reproduction, distribution, exhibition and communication at a national and international level in the different databases. data, repositories and portals.
It is hereby stated that the mentioned manuscript has not been published and that it is not being printed in any other national or foreign journal.
The authors hereby accept the necessary modifications, suggested by the reviewers, in order to adapt the manuscript to the style and publication rules of this Journal.
