Can the generalizability issue of artificial intelligence be overcome? Pneumothorax detection algorithm.

Autor:	Verdi EB; Department of Chest Diseases, Ankara University Faculty of Medicine, Ankara, Turkey., Yılmaz M; School of Engineering, TOBB University of Economics and Technology, Ankara, Turkey., Doğan Mülazimoğlu D; Department of Chest Diseases, Health Sciences University Gülhane Faculty of Medicine, Ankara, Turkey., Türker A; School of Engineering, TOBB University of Economics and Technology, Ankara, Turkey., Gürün Kaya A; Department of Chest Diseases, Ankara University Faculty of Medicine, Ankara, Turkey., Işık Ö; Department of Chest Diseases, Ankara University Faculty of Medicine, Ankara, Turkey., Bostanoğlu Karaçin A; Department of Chest Diseases, Health Sciences University Gülhane Faculty of Medicine, Ankara, Turkey., Velioğlu Yakut Ö; Department of Chest Diseases, Ankara University Faculty of Medicine, Ankara, Turkey., Yenigün BM; Department of Thoracic Surgery, Ankara University Faculty of Medicine, Ankara, Turkey., Uzun Ç; Department of Radiology, Ankara University Faculty of Medicine, Ankara, Turkey., Elhan AH; Department of Biostatistics, Ankara University Faculty of Medicine, Ankara, Turkey., Özdemir Kumbasar Ö; Department of Chest Diseases, Ankara University Faculty of Medicine, Ankara, Turkey., Kaya A; Department of Chest Diseases, Ankara University Faculty of Medicine, Ankara, Turkey., Kayı Cangır A; Department of Thoracic Surgery, Ankara University Faculty of Medicine, Ankara, Turkey., Taşçı C; Department of Chest Diseases, Health Sciences University Gülhane Faculty of Medicine, Ankara, Turkey., Özbayoğlu AM; School of Engineering, TOBB University of Economics and Technology, Ankara, Turkey., Erol S; Department of Chest Diseases, Ankara University Faculty of Medicine, Ankara, Turkey.
Jazyk:	angličtina
Zdroj:	Journal of investigative medicine : the official publication of the American Federation for Clinical Research [J Investig Med] 2024 Jan; Vol. 72 (1), pp. 88-99. Date of Electronic Publication: 2023 Nov 10.
DOI:	10.1177/10815589231208479
Abstrakt:	The generalizability of artificial intelligence (AI) models is a major issue in the field of AI applications. Therefore, we aimed to overcome the generalizability problem of an AI model developed for a particular center for pneumothorax detection using a small dataset for external validation. Chest radiographs of patients diagnosed with pneumothorax (n = 648) and those without pneumothorax (n = 650) who visited the Ankara University Faculty of Medicine (AUFM; center 1) were obtained. A deep learning-based pneumothorax detection algorithm (PDA-Alpha) was developed using the AUFM dataset. For implementation at the Health Sciences University (HSU; center 2), PDA-Beta was developed through external validation of PDA-Alpha using 50 radiographs with pneumothorax obtained from HSU. Both PDA algorithms were assessed using the HSU test dataset (n = 200) containing 50 pneumothorax and 150 non-pneumothorax radiographs. We compared the results generated by the algorithms with those of physicians to demonstrate the reliability of the results. The areas under the curve for PDA-Alpha and PDA-Beta were 0.993 (95% confidence interval (CI): 0.985-1.000) and 0.986 (95% CI: 0.962-1.000), respectively. Both algorithms successfully detected the presence of pneumothorax on 49/50 radiographs; however, PDA-Alpha had seven false-positive predictions, whereas PDA-Beta had one. The positive predictive value increased from 0.525 to 0.886 after external validation (p = 0.041). The physicians' sensitivity and specificity for detecting pneumothorax were 0.585 and 0.988, respectively. The performance scores of the algorithms were increased with a small dataset; however, further studies are required to determine the optimal amount of external validation data to fully address the generalizability issue. Competing Interests: Declaration of conflicting interestsThe author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu