{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,3]],"date-time":"2026-04-03T18:34:15Z","timestamp":1775241255798,"version":"3.50.1"},"reference-count":51,"publisher":"Oxford University Press (OUP)","issue":"3","license":[{"start":{"date-parts":[[2025,1,7]],"date-time":"2025-01-07T00:00:00Z","timestamp":1736208000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Machine Learning for Healthcare"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,3,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>The primary practice of healthcare artificial intelligence (AI) starts with model development, often using state-of-the-art AI, retrospectively evaluated using metrics lifted from the AI literature like AUROC and DICE score. However, good performance on these metrics may not translate to improved clinical outcomes. Instead, we argue for a better development pipeline constructed by working backward from the end goal of positively impacting clinically relevant outcomes using AI, leading to considerations of causality in model development and validation, and subsequently a better development pipeline. Healthcare AI should be \u201cactionable,\u201d and the change in actions induced by AI should improve outcomes. Quantifying the effect of changes in actions on outcomes is causal inference. The development, evaluation, and validation of healthcare AI should therefore account for the causal effect of intervening with the AI on clinically relevant outcomes. Using a causal lens, we make recommendations for key stakeholders at various stages of the healthcare AI pipeline. Our recommendations aim to increase the positive impact of AI on clinical outcomes.<\/jats:p>","DOI":"10.1093\/jamia\/ocae301","type":"journal-article","created":{"date-parts":[[2025,1,7]],"date-time":"2025-01-07T21:18:37Z","timestamp":1736284717000},"page":"589-594","source":"Crossref","is-referenced-by-count":16,"title":["AI as an intervention: improving clinical outcomes relies on a causal approach to AI development and validation"],"prefix":"10.1093","volume":"32","author":[{"given":"Shalmali","family":"Joshi","sequence":"first","affiliation":[{"name":"Department of Biomedical Informatics, Columbia University , New York, NY 10032,","place":["United States"]}]},{"given":"I\u00f1igo","family":"Urteaga","sequence":"additional","affiliation":[{"name":"BCAM\u2014Basque Center for Applied Mathematics , Bilbao 48009,","place":["Spain"]},{"name":"IKERBASQUE\u2014Basque Foundation for Science , Bilbao 48009,","place":["Spain"]}]},{"given":"Wouter A C","family":"van Amsterdam","sequence":"additional","affiliation":[{"name":"Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht , Utrecht 3584 CX,","place":["The Netherlands"]}]},{"given":"George","family":"Hripcsak","sequence":"additional","affiliation":[{"name":"Department of Biomedical Informatics, Columbia University , New York, NY 10032,","place":["United States"]}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9643-3024","authenticated-orcid":false,"given":"Pierre","family":"Elias","sequence":"additional","affiliation":[{"name":"Department of Biomedical Informatics, Columbia University , New York, NY 10032,","place":["United States"]},{"name":"Division of Cardiology, Columbia University , New York, NY 10032,","place":["United States"]}]},{"given":"Benjamin","family":"Recht","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering and Computer Sciences, University of California, Berkeley , Berkeley, CA 94720,","place":["United States"]}]},{"given":"No\u00e9mie","family":"Elhadad","sequence":"additional","affiliation":[{"name":"Department of Biomedical Informatics, Columbia University , New York, NY 10032,","place":["United States"]}]},{"given":"James","family":"Fackler","sequence":"additional","affiliation":[{"name":"Department of Anesthesiology and Critical Care Medicine, Johns Hopkins University School of Medicine , Baltimore, MD 21287,","place":["United States"]}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5828-4497","authenticated-orcid":false,"given":"Mark P","family":"Sendak","sequence":"additional","affiliation":[{"name":"Population Health and Data Science, Duke Institute of Health Innovation , Durham, NC 27701,","place":["United States"]}]},{"given":"Jenna","family":"Wiens","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, University of Michigan, Ann Arbor , Ann Arbor, MI 48109,","place":["United States"]}]},{"given":"Kaivalya","family":"Deshpande","sequence":"additional","affiliation":[{"name":"Department of Medicine, NYU Grossman School of Medicine , New York, NY 10016,","place":["United States"]}]},{"given":"Yoav","family":"Wald","sequence":"additional","affiliation":[{"name":"Center for Data Science, New York University , New York, NY 10011,","place":["United States"]}]},{"given":"Madalina","family":"Fiterau","sequence":"additional","affiliation":[{"name":"College of Information and Computer Sciences, University of Massachusetts, Amherst , Amherst, MA 01003,","place":["United States"]}]},{"given":"Zachary","family":"Lipton","sequence":"additional","affiliation":[{"name":"Department of Machine Learning, Carnegie Mellon University , Pittsburgh, PA 15213,","place":["United States"]}]},{"given":"Daniel","family":"Malinsky","sequence":"additional","affiliation":[{"name":"Department of Biostatistics, Columbia University , New York, NY 10032,","place":["United States"]}]},{"given":"Madhur","family":"Nayan","sequence":"additional","affiliation":[{"name":"Department of Population Health and Urology, NYU Grossman School of Medicine , New York, NY 10016,","place":["United States"]}]},{"given":"Hongseok","family":"Namkoong","sequence":"additional","affiliation":[{"name":"Division of Decisions, Risk, and Operations, Columbia Business School , New York, NY 10027,","place":["United States"]}]},{"given":"Soojin","family":"Park","sequence":"additional","affiliation":[{"name":"Department of Biomedical Informatics, Columbia University , New York, NY 10032,","place":["United States"]}]},{"given":"Julia E","family":"Vogt","sequence":"additional","affiliation":[{"name":"Department of Computer Science, ETH Zurich , Zurich 8092,","place":["Switzerland"]}]},{"given":"Rajesh","family":"Ranganath","sequence":"additional","affiliation":[{"name":"Center for Data Science, New York University , New York, NY 10011,","place":["United States"]},{"name":"Department of Computer Science, New York University , New York, NY 10012,","place":["United States"]}]}],"member":"286","published-online":{"date-parts":[[2025,1,7]]},"reference":[{"key":"2025021811392524100_ocae301-B1","doi-asserted-by":"crossref","first-page":"31","DOI":"10.1038\/s41591-021-01614-0","article-title":"AI in health and medicine","volume":"28","author":"Rajpurkar","year":"2022","journal-title":"Nat Med"},{"key":"2025021811392524100_ocae301-B2","first-page":"191","article-title":"A review of challenges and opportunities in machine learning for health","volume":"2020","author":"Ghassemi","year":"2020","journal-title":"AMIA Summits Transl Sci Proc"},{"key":"2025021811392524100_ocae301-B3","doi-asserted-by":"crossref","first-page":"393","DOI":"10.1146\/annurev-biodatasci-092820-033938","article-title":"Probabilistic machine learning for healthcare","volume":"4","author":"Chen","year":"2021","journal-title":"Annu Rev Biomed Data Sci"},{"key":"2025021811392524100_ocae301-B4","doi-asserted-by":"crossref","first-page":"58","DOI":"10.1162\/daed_a_01900","article-title":"A golden decade of deep learning: computing systems & applications","volume":"151","author":"Dean","year":"2022","journal-title":"Daedalus"},{"key":"2025021811392524100_ocae301-B5","doi-asserted-by":"crossref","first-page":"e288","DOI":"10.1016\/S2589-7500(23)00025-0","article-title":"The impact of commercial health datasets on medical research and health-care algorithms","volume":"5","author":"Alberto","year":"2023","journal-title":"Lancet Digit Health"},{"key":"2025021811392524100_ocae301-B6","doi-asserted-by":"crossref","first-page":"1337","DOI":"10.1038\/s41591-019-0548-6","article-title":"Do no harm: a roadmap for responsible machine learning for health care","volume":"25","author":"Wiens","year":"2019","journal-title":"Nat Med"},{"key":"2025021811392524100_ocae301-B7","doi-asserted-by":"crossref","first-page":"44","DOI":"10.1038\/s41591-018-0300-7","article-title":"High-performance medicine: the convergence of human and artificial intelligence","volume":"25","author":"Topol","year":"2019","journal-title":"Nat Med"},{"key":"2025021811392524100_ocae301-B8","doi-asserted-by":"crossref","first-page":"181","DOI":"10.1146\/annurev-biodatasci-080917-013343","article-title":"Deep learning in biomedical data science","volume":"1","author":"Baldi","year":"2018","journal-title":"Annu Rev Biomed Data Sci"},{"key":"2025021811392524100_ocae301-B9","doi-asserted-by":"crossref","first-page":"195","DOI":"10.1186\/s12916-019-1426-2","article-title":"Key challenges for delivering clinical impact with artificial intelligence","volume":"17","author":"Kelly","year":"2019","journal-title":"BMC Med"},{"key":"2025021811392524100_ocae301-B10","doi-asserted-by":"crossref","first-page":"77","DOI":"10.1038\/s41746-019-0155-4","article-title":"The \u201cinconvenient truth\u201d about AI in healthcare","volume":"2","author":"Panch","year":"2019","journal-title":"NPJ Digit Med"},{"key":"2025021811392524100_ocae301-B11","doi-asserted-by":"crossref","first-page":"850","DOI":"10.1186\/s12913-022-08215-8","article-title":"Challenges to implementing artificial intelligence in healthcare: a qualitative interview study with healthcare leaders in Sweden","volume":"22","author":"Petersson","year":"2022","journal-title":"BMC Health Serv Res"},{"key":"2025021811392524100_ocae301-B12","doi-asserted-by":"crossref","first-page":"565","DOI":"10.1177\/0272989X06295361","article-title":"Decision curve analysis: a novel method for evaluating prediction models","volume":"26","author":"Vickers","year":"2006","journal-title":"Med Decis Making"},{"key":"2025021811392524100_ocae301-B13","author":"van Amsterdam","year":"2023"},{"key":"2025021811392524100_ocae301-B14","doi-asserted-by":"publisher","DOI":"10.1161\/circ.148.suppl_1.14647","article-title":"Abstract 14647: EchoNext: an ECG-based deep learning model to detect structural heart disease","volume":"148","author":"Jing","year":"2023","journal-title":"Circulation"},{"key":"2025021811392524100_ocae301-B15","author":"Jethani","year":"2022"},{"key":"2025021811392524100_ocae301-B16","doi-asserted-by":"crossref","first-page":"1503","DOI":"10.1093\/jamia\/ocad088","article-title":"Implementation frameworks for end-to-end clinical AI: derivation of the SALIENT framework","volume":"30","author":"van der Vegt","year":"2023","journal-title":"J Am Med Inform Assoc"},{"key":"2025021811392524100_ocae301-B17","doi-asserted-by":"crossref","first-page":"929508","DOI":"10.3389\/fdgth.2022.929508","article-title":"The silent trial\u2014the bridge between bench-to-bedside clinical AI applications","volume":"4","author":"Kwong","year":"2022","journal-title":"Front Digit Health"},{"key":"2025021811392524100_ocae301-B18","first-page":"169","volume-title":"Proceedings of the Conference on Health, Inference, and Learning","author":"Tonekaboni","year":"2022"},{"key":"2025021811392524100_ocae301-B19","doi-asserted-by":"crossref","first-page":"7","DOI":"10.1038\/s41746-023-00753-7","article-title":"Making machine learning matter to clinicians: model actionability in medical decision-making","volume":"6","author":"Ehrmann","year":"2023","journal-title":"NPJ Digit Med"},{"key":"2025021811392524100_ocae301-B20","doi-asserted-by":"crossref","first-page":"16","DOI":"10.1038\/s41591-018-0310-5","article-title":"Guidelines for reinforcement learning in healthcare","volume":"25","author":"Gottesman","year":"2019","journal-title":"Nat Med"},{"key":"2025021811392524100_ocae301-B21","author":"Uehara","year":"2022"},{"key":"2025021811392524100_ocae301-B22","doi-asserted-by":"crossref","first-page":"3","DOI":"10.1186\/s41512-021-00092-9","article-title":"A scoping review of causal methods enabling predictions under hypothetical interventions","volume":"5","author":"Lin","year":"2021","journal-title":"Diagn Progn Res"},{"key":"2025021811392524100_ocae301-B23","doi-asserted-by":"crossref","DOI":"10.1017\/CBO9780511803161","volume-title":"Causality","author":"Pearl","year":"2009"},{"key":"2025021811392524100_ocae301-B24","volume-title":"Causal Inference","author":"Hernan","year":"2019"},{"key":"2025021811392524100_ocae301-B25","author":"Puli","year":"2022"},{"key":"2025021811392524100_ocae301-B26","author":"Ranganath","year":"2016:101-114"},{"key":"2025021811392524100_ocae301-B27","first-page":"244","article-title":"Deep survival analysis: nonparametrics and missingness","volume":"85","author":"Miscouridou","year":"2018","journal-title":"MLHC"},{"key":"2025021811392524100_ocae301-B28","article-title":"DeepHit: a deep learning approach to survival analysis with competing risks","author":"Lee","journal-title":"Proc AAAI Conf Artif Intell"},{"key":"2025021811392524100_ocae301-B29","first-page":"18296","article-title":"X-CAL: explicit calibration for survival analysis","volume":"33","author":"Goldstein","year":"2020","journal-title":"Adv Neural Inf Process Syst"},{"key":"2025021811392524100_ocae301-B30","doi-asserted-by":"crossref","first-page":"123","DOI":"10.1146\/annurev-biodatasci-092820-114757","article-title":"Ethical machine learning in healthcare","volume":"4","author":"Chen","year":"2021","journal-title":"Annu Rev Biomed Data Sci"},{"key":"2025021811392524100_ocae301-B31","volume-title":"Fairness and Machine Learning: Limitations and Opportunities","author":"Barocas","year":"2023"},{"key":"2025021811392524100_ocae301-B32","doi-asserted-by":"crossref","first-page":"866","DOI":"10.7326\/M18-1990","article-title":"Ensuring fairness in machine learning to advance health equity","volume":"169","author":"Rajkomar","year":"2018","journal-title":"Ann Intern Med"},{"key":"2025021811392524100_ocae301-B33","doi-asserted-by":"crossref","first-page":"719","DOI":"10.1038\/s41551-023-01056-8","article-title":"Algorithmic fairness in artificial intelligence for medicine and healthcare","volume":"7","author":"Chen","year":"2023","journal-title":"Nat Biomed Eng"},{"key":"2025021811392524100_ocae301-B34","doi-asserted-by":"crossref","first-page":"1728","DOI":"10.1126\/science.1150009","article-title":"Public health. Implementation science","volume":"318","author":"Madon","year":"2007","journal-title":"Science"},{"key":"2025021811392524100_ocae301-B35","first-page":"1","volume-title":"Statistical Models in Epidemiology, the Environment, and Clinical Trials","author":"Robins"},{"key":"2025021811392524100_ocae301-B36","first-page":"295","article-title":"The environment and disease: association or causation?","volume":"58","author":"Hill","year":"1965","journal-title":"Proc R Soc Med"},{"key":"2025021811392524100_ocae301-B37","doi-asserted-by":"crossref","first-page":"570","DOI":"10.1007\/s11121-012-0339-5","article-title":"An introduction to sensitivity analysis for unobserved confounding in nonexperimental prevention research","volume":"14","author":"Liu","year":"2013","journal-title":"Prev Sci"},{"key":"2025021811392524100_ocae301-B38","doi-asserted-by":"crossref","DOI":"10.3386\/w30302","volume-title":"Long Story Short: Omitted Variable Bias in Causal Machine Learning","author":"Chernozhukov","year":"2022"},{"key":"2025021811392524100_ocae301-B39","doi-asserted-by":"crossref","first-page":"29","DOI":"10.1186\/s12916-018-1017-7","article-title":"Adaptive designs in clinical trials: why use them, and how to run and report them","volume":"16","author":"Pallmann","year":"2018","journal-title":"BMC Med"},{"key":"2025021811392524100_ocae301-B40","doi-asserted-by":"crossref","first-page":"1175","DOI":"10.1056\/NEJMsb1900856","article-title":"Creating a learning health system through rapid-cycle, randomized testing","volume":"381","author":"Horwitz","year":"2019","journal-title":"N Engl J Med"},{"key":"2025021811392524100_ocae301-B41","doi-asserted-by":"crossref","first-page":"e078378","DOI":"10.1136\/bmj-2023-078378","article-title":"TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods","volume":"385","author":"Collins","year":"2024","journal-title":"BMJ"},{"key":"2025021811392524100_ocae301-B42","article-title":"Prognostic models for decision support need to report their targeted treatments and the expected changes in treatment decisions","volume":"385","author":"van Amsterdam","year":"2024","journal-title":"BMJ"},{"key":"2025021811392524100_ocae301-B43","doi-asserted-by":"crossref","first-page":"130","DOI":"10.1016\/j.crad.2022.09.131","article-title":"DECIDE-AI: a new reporting guideline and its relevance to artificial intelligence studies in radiology","volume":"78","author":"Vasey","year":"2023","journal-title":"Clin Radiol"},{"key":"2025021811392524100_ocae301-B44","doi-asserted-by":"crossref","first-page":"1364","DOI":"10.1038\/s41591-020-1034-x","article-title":"Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension","volume":"26","author":"Liu","year":"2020","journal-title":"Nat Med"},{"key":"2025021811392524100_ocae301-B45","doi-asserted-by":"crossref","first-page":"1351","DOI":"10.1038\/s41591-020-1037-7","article-title":"Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension","volume":"26","author":"Cruz Rivera","year":"2020","journal-title":"Nat Med"},{"key":"2025021811392524100_ocae301-B46","doi-asserted-by":"crossref","first-page":"639","DOI":"10.1001\/jama.2024.0549","article-title":"Regulate artificial intelligence in health care by prioritizing patient outcomes","volume":"331","author":"Ayers","year":"2024","journal-title":"JAMA"},{"key":"2025021811392524100_ocae301-B47","doi-asserted-by":"crossref","first-page":"2686","DOI":"10.1038\/s41591-023-02540-z","article-title":"External validation of AI models in health should be replaced with recurring local validation","volume":"29","author":"Youssef","year":"2023","journal-title":"Nat Med"},{"key":"2025021811392524100_ocae301-B48","doi-asserted-by":"crossref","first-page":"1532","DOI":"10.1093\/jamia\/ocad114","article-title":"DEPLOYR: a technical framework for deploying custom real-time machine learning models into the electronic medical record","volume":"30","author":"Corbin","year":"2023","journal-title":"J Am Med Inform Assoc"},{"key":"2025021811392524100_ocae301-B49","doi-asserted-by":"publisher","first-page":"41","DOI":"10.1038\/s41746-020-0253-3","article-title":"Presenting machine learning model information to clinical end users with model facts labels","volume":"3","author":"Sendak","year":"2020","journal-title":"NPJ Digit Med"},{"key":"2025021811392524100_ocae301-B50","year":"2023"},{"key":"2025021811392524100_ocae301-B51","doi-asserted-by":"crossref","first-page":"1525","DOI":"10.1038\/s41591-024-02870-6","article-title":"Off-label use of artificial intelligence models in healthcare","volume":"30","author":"Krishnamoorthy","year":"2024","journal-title":"Nat Med"}],"container-title":["Journal of the American Medical Informatics Association"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/jamia\/article-pdf\/32\/3\/589\/61371325\/ocae301.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jamia\/article-pdf\/32\/3\/589\/61371325\/ocae301.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,2,18]],"date-time":"2025-02-18T11:39:46Z","timestamp":1739878786000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/jamia\/article\/32\/3\/589\/7945189"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,1,7]]},"references-count":51,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2025,1,7]]},"published-print":{"date-parts":[[2025,3,1]]}},"URL":"https:\/\/doi.org\/10.1093\/jamia\/ocae301","relation":{},"ISSN":["1067-5027","1527-974X"],"issn-type":[{"value":"1067-5027","type":"print"},{"value":"1527-974X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2025,3]]},"published":{"date-parts":[[2025,1,7]]}}}