@article{6043ddd270cc4c56be35e65cb537cfdd,
title = "Open-Source AI Model for Predicting Respiratory Mortality in COPD from Chest Radiographs",
abstract = "Purpose: To evaluate the clinical utility of artificial intelligence (AI) scores in years derived from chest radiographs (CXR-Lung-Risk scores) in predicting respiratory mortality in a cohort of patients with chronic obstructive pulmonary disease (COPD). Materials and Methods: This retrospective single-center study included patients with COPD from a tertiary center between January 2011 and December 2015. CXR-Lung-Risk scores were derived from chest radiographs using an open-source AI algorithm. The primary outcome, respiratory mortality, was assessed for its association with CXR-Lung-Risk using a multivariable Fine-Gray model adjusted for age, sex, body mass index, smoking status, comorbidities, and pulmonary function test results. Discrimination was evaluated and benchmarked against the Global Initiative for Chronic Obstructive Lung Disease grade using the time-dependent area under the receiver operating characteristic curve (AUC). Associations between CXR-Lung-Risk and lung-function measures were examined. Results: A total of 4226 patients (median age, 70 years [IQR, 63–76 years]; 3293 male) with COPD were evaluated. Respiratory mortality was observed in 19.7\% (831 of 4226) of patients at a median follow-up of 6.7 years (IQR, 4.0–7.9 years). CXR-Lung-Risk was a prognostic factor for respiratory mortality (subdistribution hazard ratio per 5-year increase, 1.16; 95\% CI: 1.10, 1.28; P < .001) after adjusting for lung function and clinical risk factors. Likelihood-ratio testing further confirmed its added value in multivariable models (P < .001). The AUC for CXR-Lung-Risk in predicting respiratory mortality up to 10 years was 0.76 (95\% CI: 0.72, 0.79), which outperformed the Global Initiative for Chronic Obstructive Lung Disease grades (0.61; 95\% CI: 0.58, 0.65; P < .001). Pulmonary function decreased with increasing CXR-Lung-Risk scores (P < .001). Conclusion: COPD. This study demonstrates that CXR-Lung-Risk is a valuable open-source AI tool for predicting respiratory mortality among patients with.",
keywords = "Artificial Intelligence, Chest Radiograph, Chronic Obstructive Pulmonary Disease, Prognostication",
author = "Lee, \{Jong Hyuk\} and Lee, \{Chang Hoon\} and Jayoun Kim and Seungho Lee and Jakob Weiss and Raghu, \{Vineet K.\} and Lu, \{Michael T.\} and Aerts, \{Hugo J.W.L.\} and Kang, \{Hye Rin\} and Nam, \{Ju Gang\} and Park, \{Chang Min\} and Goo, \{Jin Mo\} and Hyungjin Kim",
note = "Funding Information: Disclosures of conflicts of interest: J.H.L. Consulting fees from RadiSen; a research grant from Coreline Soft. C.H.L. Research grant from GlaxoSmithKline. J.K. No relevant relationships. S.L. No relevant relationships. J.W. German Research Foundation grant, project number: 525002713; grant from Siemens Healthineers; payment or honoraria from Onc.AI. V.K.R. Grants to institution from American Heart Association, National Heart, Lung, and Blood Institute, Norn Group, Johnson and Johnson Innovation, National Academy of Medicine; payment for lecture from Boehringer Ingelheim; stock in Apple, Amazon, Alphabet, NVIDIA, and Meta. M.T.L. Research funding to institution from American Heart Association, AstraZeneca, Ionis, Johnson \& Johnson Innovation, Kowa, MedImmune, NHLBI, National Academy of Medicine, Risk Management Foundation of the Harvard Medical Institutions; consulting fees from Eli Lilly. H.J.W.L.A. Scientific advisor and/or shareholder of Onc.AI, Love Health, Ambient, Health-AI, Sphera, Editas, AZ, and BMS (outside of presented work). H.R.K. No relevant relationships. J.G.N. Research grant from Vuno. C.M.P. Board member of Korean Society of Radiology, Korean Society of Thoracic Radiology, and Korean Society of Artificial Intelligence in Medicine; stock in Promedius and Monitor; stock options in Lunit and Coreline Soft; research grants from Lunit, Coreline Soft, and HealthHub. J.M.G. Research grants from Coreline Soft and Taejoon Pharm. H.K. National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (no. RS-2023-00207978) to institution; research grants from Kakao Brain and RadiSen to institution; consulting fees from Radisen to author and institution; stock and stock options in Medical IP; stock in Soombit.ai; employed as Medical Director at Soombit.ai; member of Radiology Advances editorial board; an honorarium from AstraZeneca. Funding Information: Supported by a National Research Foundation of Korea grant funded by the Korean government (Ministry of Science and ICT) (no. RS-2023-00207978). However, the funders had no role in the study design; in the collection, analysis, and interpretation of the data; in the writing of the report; and in the decision to submit the article for publication. Funding Information: Funding: Supported by a National Research Foundation of Korea grant funded by the Korean government (Ministry of Science and ICT) (no. RS-2023-00207978). However, the funders had no role in the study design; in the collection, analysis, and interpretation of the data; in the writing of the report; and in the decision to submit the article for publication. Publisher Copyright: {\textcopyright} RSNA, 2025.",
year = "2025",
month = oct,
day = "1",
doi = "10.1148/ryct.250080",
language = "English",
volume = "7",
journal = "Radiology. Cardiothoracic imaging",
issn = "2638-6135",
publisher = "Radiological Society of North America Inc.",
number = "5",
}