{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,19]],"date-time":"2026-11-19T19:17:16Z","timestamp":1795115836321,"version":"4.10.3"},"reference-count":44,"publisher":"Public Library of Science (PLoS)","issue":"11","license":[{"start":{"date-parts":[[2026,11,19]],"date-time":"2026-11-19T00:00:00Z","timestamp":1795046400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100016275","name":"Seoul National University Bundang Hospital","doi-asserted-by":"publisher","award":["22-2024-0008"],"award-info":[{"award-number":["22-2024-0008"]}],"id":[{"id":"10.13039\/100016275","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100016275","name":"Seoul National University Bundang Hospital","doi-asserted-by":"publisher","award":["09-2024-0002"],"award-info":[{"award-number":["09-2024-0002"]}],"id":[{"id":"10.13039\/100016275","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["www.plosdigitalhealth.org"],"crossmark-restriction":false},"short-container-title":["PLOS Digit Health"],"abstract":"<jats:p>Frailty is a critical age-related syndrome associated with increased risks of disability, hospitalization, and mortality. While Comprehensive Geriatric Assessment (CGA) is the standard for evaluation, it is often subjective, time-consuming, and resource-intensive. To address these limitations, this study developed and validated a multimodal deep learning framework that integrates chest X-rays (CXR) and electrocardiograms (ECG) to provide an objective and rapid tool for frailty classification. We retrospectively analyzed data from 2,830 older adults who underwent CGA between 2007 and 2024, partitioned chronologically into training, validation, and test sets. Frailty was defined using a Multidimensional Frailty Score (MFS) &gt; 5. Three deep learning models were trained and evaluated: a CXR-only model (EVA-X vision transformer), an ECG-only model (DenseNet-121), and a multimodal model combining both inputs. The multimodal model outperformed the single-modality models and four non-imaging clinical baseline models incorporating age, sex, comorbidity index, or routine laboratory variables, achieving the highest area under the receiver operating characteristic curve (AUROC) of 0.778 and area under the precision-recall curve (AUPRC) of 0.726. For continuous MFS estimation, the multimodal model achieved the highest Pearson correlation (r\u2009=\u20090.610) and the lowest mean absolute error (MAE\u2009=\u20091.68) among the evaluated models. Gradient-weighted Class Activation Mapping (Grad-CAM) was used to qualitatively explore the image regions that influenced model predictions. This multimodal AI framework demonstrates modest classification performance and feasibility for frailty screening. By leveraging routinely collected clinical images, this framework may serve as a practical screening adjunct to identify older adults who may benefit from formal CGA. Further studies are needed to determine its clinical utility when applied preoperatively for the stratification of postoperative risk.<\/jats:p>","DOI":"10.1371\/journal.pdig.0001838","type":"journal-article","created":{"date-parts":[[2026,11,19]],"date-time":"2026-11-19T18:38:18Z","timestamp":1795113498000},"page":"e0001838","update-policy":"https:\/\/doi.org\/10.1371\/journal.pdig.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["Multimodal deep learning approach for frailty classification using electrocardiogram and chest X-ray imaging"],"prefix":"10.1371","volume":"5","author":[{"given":"JinYong","family":"Park","sequence":"first","affiliation":[{"name":"Seoul National University Bundang Hospital","place":["Seongnam-si, Republic of Korea"],"department":["Center for Artificial Intelligence in Healthcare"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5139-5346","authenticated-orcid":true,"given":"Jung-Yeon","family":"Choi","sequence":"additional","affiliation":[{"name":"Seoul National University Bundang Hospital","place":["Seongnam-si, Republic of Korea"],"department":["Department of Internal Medicine"]},{"name":"Seoul National University College of Medicine","place":["Seoul, Republic of Korea"],"department":["Department of Internal Medicine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6658-047X","authenticated-orcid":true,"given":"Kwang-il","family":"Kim","sequence":"additional","affiliation":[{"name":"Seoul National University Bundang Hospital","place":["Seongnam-si, Republic of Korea"],"department":["Department of Internal Medicine"]},{"name":"Seoul National University College of Medicine","place":["Seoul, Republic of Korea"],"department":["Department of Internal Medicine"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,11,19]]},"reference":[{"issue":"1","key":"pdig.0001838.ref001","doi-asserted-by":"crossref","first-page":"25","DOI":"10.1016\/j.cger.2017.09.004","article-title":"Frailty screening and interventions: Considerations for clinical practice","volume":"34","author":"J Walston","year":"2018","journal-title":"Clin Geriatr Med"},{"issue":"3","key":"pdig.0001838.ref002","doi-asserted-by":"crossref","DOI":"10.1093\/gerona\/56.3.M146","article-title":"Frailty in older adults: Evidence for a phenotype","volume":"56","author":"LP Fried","year":"2001","journal-title":"J Gerontol A Biol Sci Med Sci"},{"issue":"9868","key":"pdig.0001838.ref003","doi-asserted-by":"crossref","first-page":"752","DOI":"10.1016\/S0140-6736(12)62167-9","article-title":"Frailty in elderly people","volume":"381","author":"A Clegg","year":"2013","journal-title":"Lancet"},{"key":"pdig.0001838.ref004","doi-asserted-by":"crossref","first-page":"1065779","DOI":"10.3389\/fcvm.2023.1065779","article-title":"Frailty score and outcomes of patients undergoing vascular surgery and amputation: A systematic review and meta-analysis","volume":"10","author":"S Chen","year":"2023","journal-title":"Front Cardiovasc Med"},{"issue":"6","key":"pdig.0001838.ref005","doi-asserted-by":"crossref","first-page":"538","DOI":"10.1056\/NEJMra2301292","article-title":"Frailty in older adults","volume":"391","author":"DH Kim","year":"2024","journal-title":"N Engl J Med"},{"issue":"4","key":"pdig.0001838.ref006","doi-asserted-by":"crossref","first-page":"681","DOI":"10.1111\/j.1532-5415.2010.02764.x","article-title":"Prevalence and 10-year outcomes of frailty in older adults in relation to deficit accumulation","volume":"58","author":"X Song","year":"2010","journal-title":"J Am Geriatr Soc"},{"key":"pdig.0001838.ref007","doi-asserted-by":"crossref","first-page":"55","DOI":"10.1186\/s12877-016-0225-2","article-title":"Psychometric properties of multicomponent tools designed to assess frailty in older adults: A systematic review","volume":"16","author":"JL Sutton","year":"2016","journal-title":"BMC Geriatr"},{"issue":"1","key":"pdig.0001838.ref008","doi-asserted-by":"crossref","first-page":"96","DOI":"10.1093\/ageing\/afaa219","article-title":"Prevalence of frailty in 62 countries across the world: A systematic review and meta-analysis of population-level studies","volume":"50","author":"R O\u2019Caoimh","year":"2021","journal-title":"Age Ageing"},{"issue":"3","key":"pdig.0001838.ref009","doi-asserted-by":"crossref","first-page":"1","DOI":"10.4017\/gt.2020.19.003.06","article-title":"Technology for home-based frailty assessment and prediction: A systematic review","volume":"19","author":"C Bian","year":"2020","journal-title":"Gerontechnology"},{"issue":"3","key":"pdig.0001838.ref010","doi-asserted-by":"crossref","first-page":"262","DOI":"10.1093\/gerona\/61.3.262","article-title":"Phenotype of frailty: Characterization in the women\u2019s health and aging studies","volume":"61","author":"K Bandeen-Roche","year":"2006","journal-title":"J Gerontol A Biol Sci Med Sci"},{"issue":"5","key":"pdig.0001838.ref011","doi-asserted-by":"crossref","DOI":"10.1093\/ageing\/afac104","article-title":"Comprehensive geriatric assessment in older people: An umbrella review of health outcomes","volume":"51","author":"N Veronese","year":"2022","journal-title":"Age Ageing"},{"issue":"12","key":"pdig.0001838.ref012","doi-asserted-by":"crossref","first-page":"721","DOI":"10.1136\/bmjqs-2022-015832","article-title":"Did the Acute Frailty Network improve outcomes for older people living with frailty? A staggered difference-in-difference panel event study","volume":"32","author":"A Street","year":"2023","journal-title":"BMJ Qual Saf"},{"issue":"7","key":"pdig.0001838.ref013","article-title":"Deep learning to assess long-term mortality from chest radiographs","volume":"2","author":"MT Lu","year":"2019","journal-title":"JAMA Netw Open"},{"issue":"11","key":"pdig.0001838.ref014","doi-asserted-by":"crossref","first-page":"2226","DOI":"10.1016\/j.jcmg.2021.01.008","article-title":"Deep learning to estimate biological age from chest radiographs","volume":"14","author":"VK Raghu","year":"2021","journal-title":"JACC Cardiovasc Imaging"},{"issue":"9","key":"pdig.0001838.ref015","doi-asserted-by":"crossref","DOI":"10.1161\/CIRCEP.119.007284","article-title":"Age and sex estimation using artificial intelligence from standard 12-lead ECGs","volume":"12","author":"ZI Attia","year":"2019","journal-title":"Circ Arrhythm Electrophysiol"},{"issue":"6","key":"pdig.0001838.ref016","doi-asserted-by":"crossref","first-page":"886","DOI":"10.1038\/s41591-020-0870-z","article-title":"Prediction of mortality from 12-lead electrocardiogram voltage data using a deep neural network","volume":"26","author":"S Raghunath","year":"2020","journal-title":"Nat Med"},{"issue":"4","key":"pdig.0001838.ref017","doi-asserted-by":"crossref","first-page":"456","DOI":"10.1002\/agm2.12353","article-title":"Short-term heart rate variability: A potential approach to frailty assessment in older adults","volume":"7","author":"GC Dewangan","year":"2024","journal-title":"Aging Med (Milton)"},{"issue":"1","key":"pdig.0001838.ref018","doi-asserted-by":"crossref","first-page":"322","DOI":"10.1038\/s41746-025-01727-7","article-title":"Determinants of artificial intelligence electrocardiogram-derived age and its association with cardiovascular events and mortality: A systematic review and meta-analysis","volume":"8","author":"S Mossavarali","year":"2025","journal-title":"NPJ Digit Med"},{"issue":"11","key":"pdig.0001838.ref019","doi-asserted-by":"crossref","first-page":"781","DOI":"10.2459\/JCM.0000000000001670","article-title":"Image-based ECG analyzing deep-learning algorithm to predict biological age and mortality risks: Interethnic validation","volume":"25","author":"Y Cho","year":"2024","journal-title":"J Cardiovasc Med (Hagerstown)"},{"key":"pdig.0001838.ref020","doi-asserted-by":"crossref","first-page":"105172","DOI":"10.1016\/j.ijmedinf.2023.105172","article-title":"Machine learning approaches for frailty detection, prediction and classification in elderly people: A systematic review","volume":"178","author":"M Leghissa","year":"2023","journal-title":"Int J Med Inform"},{"key":"pdig.0001838.ref021","doi-asserted-by":"crossref","first-page":"102529","DOI":"10.1016\/j.arr.2024.102529","article-title":"Advances of artificial intelligence in predicting frailty using real-world data: A scoping review","volume":"101","author":"C Bai","year":"2024","journal-title":"Ageing Res Rev"},{"issue":"7","key":"pdig.0001838.ref022","doi-asserted-by":"crossref","first-page":"3409","DOI":"10.1109\/JBHI.2022.3152538","article-title":"Frailty identification using heart rate dynamics: A deep learning approach","volume":"26","author":"M Eskandari","year":"2022","journal-title":"IEEE J Biomed Health Inform"},{"key":"pdig.0001838.ref023","doi-asserted-by":"crossref","first-page":"269","DOI":"10.1097\/CPT.0000000000000296","article-title":"Machine learning to predict patients at risk for frailty in postoperative cardiac patients using electrocardiogram and accelerometer data","volume":"36","author":"C Jenks","year":"2025","journal-title":"Cardiopulm Phys Ther J"},{"issue":"4","key":"pdig.0001838.ref024","first-page":"1825","article-title":"Wearable-based assessment of heart rate response to physical stressors in patients after open-heart surgery with frailty","volume":"27","author":"D Sokas","year":"2023","journal-title":"IEEE J Biomed Health Inform"},{"issue":"4","key":"pdig.0001838.ref025","doi-asserted-by":"crossref","DOI":"10.1111\/eci.70194","article-title":"Exploring heart rate variability as a biomarker for frailty: A narrative review","volume":"56","author":"L Tagliafico","year":"2026","journal-title":"Eur J Clin Invest"},{"issue":"6","key":"pdig.0001838.ref026","doi-asserted-by":"crossref","first-page":"745","DOI":"10.1016\/j.hlc.2025.12.017","article-title":"Wearable cardiac devices as windows into physiological decline: A review of digital biomarkers in the prevention, detection, and management of cardiogeriatric frailty","volume":"35","author":"AA Kumar","year":"2026","journal-title":"Heart Lung Circ"},{"issue":"1","key":"pdig.0001838.ref027","doi-asserted-by":"crossref","first-page":"418","DOI":"10.1002\/jcsm.13144","article-title":"Chest X-ray-based opportunistic screening of sarcopenia using deep learning","volume":"14","author":"J Ryu","year":"2023","journal-title":"J Cachexia Sarcopenia Muscle"},{"issue":"5","key":"pdig.0001838.ref028","doi-asserted-by":"crossref","DOI":"10.1093\/radadv\/umaf035","article-title":"XComposition: Multimodal deep learning model to measure body composition using chest radiographs and clinical data","volume":"2","author":"E Alipour","year":"2025","journal-title":"Radiol Adv"},{"issue":"1","key":"pdig.0001838.ref029","doi-asserted-by":"crossref","first-page":"589","DOI":"10.1186\/s12967-026-08079-0","article-title":"Transformer-based multimodal model for estimation of appendicular lean mass using incomplete chest radiographs and electronic health record","volume":"24","author":"K Kita","year":"2026","journal-title":"J Transl Med"},{"issue":"10","key":"pdig.0001838.ref030","doi-asserted-by":"crossref","DOI":"10.1093\/gerona\/glaf173","article-title":"Deep learning chest X-ray age, epigenetic aging clocks, and associations with age-related subclinical disease in the Project Baseline Health Study","volume":"80","author":"J Chandra","year":"2025","journal-title":"J Gerontol A Biol Sci Med Sci"},{"issue":"7","key":"pdig.0001838.ref031","doi-asserted-by":"crossref","first-page":"633","DOI":"10.1001\/jamasurg.2014.241","article-title":"Multidimensional frailty score for the prediction of postoperative mortality risk","volume":"149","author":"S Kim","year":"2014","journal-title":"JAMA Surg"},{"issue":"1","key":"pdig.0001838.ref032","doi-asserted-by":"crossref","first-page":"234","DOI":"10.1186\/s12877-021-02150-9","article-title":"How does the multidimensional frailty score compare with grip strength for predicting outcomes after hip fracture surgery in older patients? A retrospective cohort study","volume":"21","author":"J-Y Choi","year":"2021","journal-title":"BMC Geriatr"},{"issue":"2","key":"pdig.0001838.ref033","doi-asserted-by":"crossref","first-page":"432","DOI":"10.1002\/jcsm.12509","article-title":"Comparison of multidimensional frailty score, grip strength, and gait speed in older surgical patients","volume":"11","author":"J-Y Choi","year":"2020","journal-title":"J Cachexia Sarcopenia Muscle"},{"key":"pdig.0001838.ref034","doi-asserted-by":"crossref","first-page":"42966","DOI":"10.1038\/srep42966","article-title":"Prediction of mortality and postoperative complications using the hip-multidimensional frailty score in elderly patients with hip fracture","volume":"7","author":"J-Y Choi","year":"2017","journal-title":"Sci Rep"},{"issue":"3","key":"pdig.0001838.ref035","article-title":"Prediction of postoperative complications using multidimensional frailty score in older female cancer patients with American Society of Anesthesiologists physical status class 1 or 2","volume":"221","author":"J-Y Choi","year":"2015","journal-title":"J Am Coll Surg"},{"key":"pdig.0001838.ref036","doi-asserted-by":"crossref","unstructured":"Chen T, Guestrin C. XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: Association for Computing Machinery; 2016. p. 785-\u201394.","DOI":"10.1145\/2939672.2939785"},{"key":"pdig.0001838.ref037","author":"JaidedAI"},{"issue":"1","key":"pdig.0001838.ref038","doi-asserted-by":"crossref","first-page":"678","DOI":"10.1038\/s41746-025-02032-z","article-title":"EVA-X: A foundation model for general chest X-ray analysis with self-supervised learning","volume":"8","author":"J Yao","year":"2025","journal-title":"NPJ Digit Med"},{"issue":"1","key":"pdig.0001838.ref039","doi-asserted-by":"crossref","first-page":"159","DOI":"10.1038\/s43856-022-00220-6","article-title":"Deep learning-based age estimation from chest X-rays indicates cardiovascular prognosis","volume":"2","author":"H Ieki","year":"2022","journal-title":"Commun Med (Lond)"},{"key":"pdig.0001838.ref040","doi-asserted-by":"crossref","unstructured":"Huang G, Liu Z, van der Maaten L, Weinberger KQ. Densely connected convolutional networks. Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2017 Jul 21\u201326; Honolulu, HI, USA. IEEE; 2017. p. 2261\u20139.","DOI":"10.1109\/CVPR.2017.243"},{"key":"pdig.0001838.ref041","unstructured":"Loshchilov I, Hutter F. Decoupled weight decay regularization. 7th International Conference on Learning Representations (ICLR 2019); 2019 May 6\u20139; New Orleans, LA, USA."},{"key":"pdig.0001838.ref042","unstructured":"Paszke A, Gross S, Massa F, Lerer A, Bradbury J, Chanan G, et al. PyTorch: An imperative style, high-performance deep learning library. Advances in Neural Information Processing Systems 32 (NeurIPS 2019); 2019 Dec 8\u201314; Vancouver, Canada. Curran Associates; 2019. p. 8024\u201335."},{"issue":"3","key":"pdig.0001838.ref043","doi-asserted-by":"crossref","first-page":"837","DOI":"10.2307\/2531595","article-title":"Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach","volume":"44","author":"ER DeLong","year":"1988","journal-title":"Biometrics"},{"key":"pdig.0001838.ref044","doi-asserted-by":"crossref","first-page":"336","DOI":"10.1007\/s11263-019-01228-7","article-title":"Grad-CAM: Visual explanations from deep networks via gradient-based localization","volume":"128","author":"RR Selvaraju","year":"2020","journal-title":"Int J Comput Vis"}],"container-title":["PLOS Digital Health"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pdig.0001838","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,11,19]],"date-time":"2026-11-19T18:38:25Z","timestamp":1795113505000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pdig.0001838"}},"subtitle":[],"editor":[{"given":"Ziyue","family":"Chen","sequence":"first","affiliation":[{"name":"Peking University","place":["CHINA"]}],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,11,19]]},"references-count":44,"journal-issue":{"issue":"11","published-online":{"date-parts":[[2026,11,19]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838","relation":{"has-part":[{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.g001","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.g002","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.g003","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.g004","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.g005","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.g006","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.t001","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.t002","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.s001","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.s002","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.s003","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.s004","asserted-by":"subject"},{"id-type":"doi","id":"https:\/\/doi.org\/10.1371\/journal.pdig.0001838.s005","asserted-by":"subject"}]},"ISSN":["2767-3170"],"issn-type":[{"value":"2767-3170","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,11,19]]}}}