{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,3]],"date-time":"2026-09-03T13:14:52Z","timestamp":1788441292087,"version":"build-2803163510"},"reference-count":55,"publisher":"Wiley","license":[{"start":{"date-parts":[[2026,9,3]],"date-time":"2026-09-03T00:00:00Z","timestamp":1788393600000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"},{"start":{"date-parts":[[2026,9,3]],"date-time":"2026-09-03T00:00:00Z","timestamp":1788393600000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/doi.wiley.com\/10.1002\/tdm_license_1.1"}],"content-domain":{"domain":["ascpt.onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Clin Pharma and Therapeutics"],"abstract":"<jats:p>\n                    Machine learning (\n                    <jats:styled-content style=\"fixed-case\">ML<\/jats:styled-content>\n                    ) is increasingly utilized in medical research. However, underrecognized methodological pitfalls can result in misleading conclusions, which if acted upon can result in patient mismanagement and harm. As a result, we have provided a compilation of common pitfalls and practical safeguard considerations for\n                    <jats:styled-content style=\"fixed-case\">ML<\/jats:styled-content>\n                    \u2010enabled medical research. Examples emphasize the importance of integrating clinical\/biologic knowledge and statistical\/epidemiologic principles, illustrated with key case examples. Topics include (1) Data drift, which can create artificial trends when coding systems or clinical practices change over time; (2) Feature construction that should be clinically grounded, as models may otherwise learn operational proxies rather than true biology; (3) Accidental data leakage, which can inflate apparent model performance and collapse on real\u2010world deployment; (4) Competing risks and differential time\u2010at\u2010risk that can generate spurious associations in safety analyses if not modeled explicitly; (5) Confounding and collider bias, which can distort findings, particularly in\n                    <jats:styled-content style=\"fixed-case\">EHR<\/jats:styled-content>\n                    \u2010based studies; (6) For rare outcomes, inappropriate evaluation measures can overstate performance and mislead explainability analyses; (7) Finally, pooled multi\u2010trial analyses require careful handling of the intrinsic heterogeneity to avoid extracting study artifacts rather than generalizable signals. By outlining many of the common potential issues, with lessons learned, we aim to provide the medical and research community with practical insights to help foster trustworthy\n                    <jats:styled-content style=\"fixed-case\">ML<\/jats:styled-content>\n                    \u2010enabled medical research.\n                  <\/jats:p>","DOI":"10.1002\/cpt.70469","type":"journal-article","created":{"date-parts":[[2026,9,3]],"date-time":"2026-09-03T12:24:36Z","timestamp":1788438276000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["<scp>AI<\/scp>\n                    Cautionary Guide: Pitfalls and Strategies for the Use of Machine Learning in Medical Research"],"prefix":"10.1002","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-4053-4213","authenticated-orcid":false,"given":"Qi","family":"Liu","sequence":"first","affiliation":[{"name":"Office of Translational Science Center for Drug Evaluation and Research, U.S. Food and Drug Administration  Silver Spring Maryland USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5806-4961","authenticated-orcid":false,"given":"Ruihao","family":"Huang","sequence":"additional","affiliation":[{"name":"Office of Translational Science Center for Drug Evaluation and Research, U.S. Food and Drug Administration  Silver Spring Maryland USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hanrui","family":"Zhang","sequence":"additional","affiliation":[{"name":"Office of Translational Science Center for Drug Evaluation and Research, U.S. Food and Drug Administration  Silver Spring Maryland USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8896-2487","authenticated-orcid":false,"given":"Thomas F.","family":"Osborne","sequence":"additional","affiliation":[{"name":"Department of Neurosurgery Stanford University School of Medicine  Palo Alto California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2026,9,3]]},"reference":[{"key":"e_1_2_8_1_1","doi-asserted-by":"crossref","first-page":"1347","DOI":"10.1056\/NEJMra1814259","article-title":"Machine learning in medicine","volume":"380","author":"Rajkomar A.","year":"2019","journal-title":"N. Engl. J. Med."},{"key":"e_1_2_8_2_1","doi-asserted-by":"crossref","first-page":"1920","DOI":"10.1161\/CIRCULATIONAHA.115.001593","article-title":"Machine learning in medicine","volume":"132","author":"Deo R.C.","year":"2015","journal-title":"Circulation"},{"key":"e_1_2_8_3_1","doi-asserted-by":"crossref","DOI":"10.1016\/j.cosrev.2021.100370","article-title":"Role of machine learning in medical research: a survey","volume":"40","author":"Garg A.","year":"2021","journal-title":"Comput Sci Rev"},{"key":"e_1_2_8_4_1","doi-asserted-by":"crossref","first-page":"232","DOI":"10.1186\/s12859-022-04753-4","article-title":"Empowering the discovery of novel target\u2010disease associations via machine learning approaches in the open targets platform","volume":"23","author":"Han Y.","year":"2022","journal-title":"BMC Bioinformatics."},{"key":"e_1_2_8_5_1","doi-asserted-by":"crossref","first-page":"748","DOI":"10.1016\/j.drudis.2020.03.003","article-title":"Machine learning models for drug\u2013target interactions: current knowledge and future directions","volume":"25","author":"D'Souza S.","year":"2020","journal-title":"Drug Discov. Today"},{"key":"e_1_2_8_6_1","doi-asserted-by":"crossref","first-page":"8891","DOI":"10.1038\/s41467-024-53333-y","article-title":"Expanding drug targets for 112 chronic diseases using a machine learning\u2010assisted genetic priority score","volume":"15","author":"Chen R.","year":"2024","journal-title":"Nat. Commun."},{"key":"e_1_2_8_7_1","doi-asserted-by":"crossref","DOI":"10.1038\/s41598-019-49656-2","article-title":"Machine learning for comprehensive forecasting of Alzheimer's disease progression","volume":"9","author":"Fisher C.K.","year":"2019","journal-title":"Sci. Rep."},{"key":"e_1_2_8_8_1","doi-asserted-by":"crossref","first-page":"673","DOI":"10.1002\/cpt.3152","article-title":"Current status and future directions: the application of artificial intelligence\/machine learning for precision medicine","volume":"115","author":"Naik K.","year":"2024","journal-title":"Clin. Pharmacol. Therap."},{"key":"e_1_2_8_9_1","doi-asserted-by":"crossref","first-page":"e677","DOI":"10.1016\/S2589-7500(20)30200-4","article-title":"Time to reality check the promises of machine learning\u2010powered precision medicine","volume":"2","author":"Wilkinson J.","year":"2020","journal-title":"Lancet Digit. Health"},{"key":"e_1_2_8_10_1","first-page":"57","volume-title":"Progress in Molecular Biology and Translational Science","author":"Sahu M.","year":"2022"},{"key":"e_1_2_8_11_1","doi-asserted-by":"crossref","first-page":"2253","DOI":"10.1016\/j.arth.2019.05.021","article-title":"Remote patient monitoring using Mobile health for Total knee arthroplasty: validation of a wearable and machine learning\u2013based surveillance platform","volume":"34","author":"Ramkumar P.N.","year":"2019","journal-title":"J. Arthroplasty"},{"key":"e_1_2_8_12_1","doi-asserted-by":"crossref","DOI":"10.1002\/widm.1485","article-title":"Remote patient monitoring using artificial intelligence: current state, applications, and challenges","volume":"13","author":"Shaik T.","year":"2023","journal-title":"WIREs Data Mining and Knowledge Discovery."},{"key":"e_1_2_8_13_1","doi-asserted-by":"crossref","first-page":"75","DOI":"10.1038\/s41569-020-00445-9","article-title":"Integration of novel monitoring devices with machine learning technology for scalable cardiovascular management","volume":"18","author":"Krittanawong C.","year":"2021","journal-title":"Nat. Rev. Cardiol."},{"key":"e_1_2_8_14_1","doi-asserted-by":"crossref","first-page":"1317","DOI":"10.1001\/jama.2017.18391","article-title":"Big data and machine learning in health care","volume":"319","author":"Beam A.L.","year":"2018","journal-title":"JAMA"},{"key":"e_1_2_8_15_1","doi-asserted-by":"crossref","DOI":"10.1259\/bjr.20220878","article-title":"Data drift in medical machine learning: implications and potential remedies","volume":"96","author":"Sahiner B.","year":"2023","journal-title":"Br. J. Radiol."},{"key":"e_1_2_8_16_1","doi-asserted-by":"crossref","first-page":"283","DOI":"10.1056\/NEJMc2104626","article-title":"The clinician and dataset shift in artificial intelligence","volume":"385","author":"Finlayson S.G.","year":"2021","journal-title":"N. Engl. J. Med."},{"key":"e_1_2_8_17_1","doi-asserted-by":"crossref","DOI":"10.1016\/j.joclim.2023.100256","article-title":"Trends in heat related illness: Nationwide observational cohort at the US Department of veteran affairs","volume":"12","author":"Osborne T.","year":"2023","journal-title":"J. Clim. Change Health"},{"key":"e_1_2_8_18_1","doi-asserted-by":"crossref","first-page":"e2513685","DOI":"10.1001\/jamanetworkopen.2025.13685","article-title":"Detecting and remediating harmful data shifts for the responsible deployment of clinical AI models","volume":"8","author":"Subasri V.","year":"2025","journal-title":"JAMA Netw. Open"},{"key":"e_1_2_8_19_1","unstructured":"D'AmourAet al.Underspecification Presents Challenges for Credibility in Modern Machine Learning. ArXiv(2020)."},{"key":"e_1_2_8_20_1","doi-asserted-by":"crossref","first-page":"610","DOI":"10.1038\/s42256-021-00338-7","article-title":"AI for radiographic COVID\u201019 detection selects shortcuts over signal","volume":"3","author":"DeGrave A.J.","year":"2021","journal-title":"Nat. Mach. Intell."},{"key":"e_1_2_8_21_1","doi-asserted-by":"crossref","first-page":"4423","DOI":"10.1038\/s41467-021-24698-1","article-title":"The impact of site\u2010specific digital histology signatures on deep learning model accuracy and bias","volume":"12","author":"Howard F.M.","year":"2021","journal-title":"Nat. Commun."},{"key":"e_1_2_8_22_1","doi-asserted-by":"crossref","first-page":"1135","DOI":"10.1001\/jamadermatol.2019.1735","article-title":"Association between surgical skin markings in Dermoscopic images and diagnostic performance of a deep learning convolutional neural network for melanoma recognition","volume":"155","author":"Winkler J.K.","year":"2019","journal-title":"JAMA Dermatol."},{"key":"e_1_2_8_23_1","article-title":"Caution regarding the specificities of pan\u2010cancer microbial structure","volume":"9","author":"Gihawi A.","year":"2023","journal-title":"Microb Genom."},{"key":"e_1_2_8_24_1","doi-asserted-by":"crossref","DOI":"10.1136\/bmj.k1479","article-title":"Biases in electronic health record data due to processes within the healthcare system: retrospective observational study","volume":"361","author":"Agniel D.","year":"2018","journal-title":"BMJ"},{"key":"e_1_2_8_25_1","doi-asserted-by":"crossref","first-page":"18","DOI":"10.1038\/s41746-020-0226-6","article-title":"Investigating sources of inaccuracy in wearable optical heart rate sensors","volume":"3","author":"Bent B.","year":"2020","journal-title":"NPJ Digit. Med."},{"key":"e_1_2_8_26_1","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pmed.1002683","article-title":"Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: a cross\u2010sectional study","volume":"15","author":"Zech J.R.","year":"2018","journal-title":"PLoS Med."},{"key":"e_1_2_8_27_1","volume-title":"Letter of Authorization: Emergency Use Authorization for Kineret (anakinra)","author":"Hinton D.M.","year":"2022"},{"key":"e_1_2_8_28_1","doi-asserted-by":"crossref","first-page":"890","DOI":"10.1002\/cpt.3191","article-title":"Using machine learning to determine a suitable patient population for Anakinra for the treatment of COVID\u201019 under the emergency use authorization","volume":"115","author":"Liu Q.","year":"2024","journal-title":"Clin. Pharmacol. Ther."},{"key":"e_1_2_8_29_1","doi-asserted-by":"crossref","DOI":"10.1093\/jnci\/dju124","article-title":"Prognostic role of neutrophil\u2010to\u2010lymphocyte ratio in solid tumors: a systematic review and meta\u2010analysis","volume":"106","author":"Templeton A.J.","year":"2014","journal-title":"J. Natl. Cancer Inst."},{"key":"e_1_2_8_30_1","doi-asserted-by":"crossref","DOI":"10.1016\/j.patter.2023.100804","article-title":"Leakage and the reproducibility crisis in machine\u2010learning\u2010based science","volume":"4","author":"Kapoor S.","year":"2023","journal-title":"Patterns (N Y)."},{"key":"e_1_2_8_31_1","unstructured":"precisionFDA.The Veterans Cardiac Health and AI Model Predictions (V\u2010CHAMPS) Challenge 2023<https:\/\/precision.fda.gov\/challenges\/31>."},{"key":"e_1_2_8_32_1","doi-asserted-by":"crossref","first-page":"e2550454","DOI":"10.1001\/jamanetworkopen.2025.50454","article-title":"Diagnostic codes in AI prediction models and label leakage of same\u2010admission clinical outcomes","volume":"8","author":"Ramadan B.","year":"2025","journal-title":"JAMA Netw. Open"},{"key":"e_1_2_8_33_1","doi-asserted-by":"crossref","first-page":"658","DOI":"10.1002\/bdra.23271","article-title":"Immortal time bias in observational studies of drug effects in pregnancy","volume":"100","author":"Matok I.","year":"2014","journal-title":"Birth Defects Res. A Clin. Mol. Teratol."},{"key":"e_1_2_8_34_1","doi-asserted-by":"crossref","first-page":"601","DOI":"10.1161\/CIRCULATIONAHA.115.017719","article-title":"Introduction to the analysis of survival data in the presence of competing risks","volume":"133","author":"Austin P.C.","year":"2016","journal-title":"Circulation"},{"key":"e_1_2_8_35_1","doi-asserted-by":"crossref","first-page":"496","DOI":"10.1080\/01621459.1999.10474144","article-title":"A proportional hazards model for the subdistribution of a competing risk","volume":"94","author":"Fine J.P.","year":"1999","journal-title":"J. Am. Stat. Assoc."},{"key":"e_1_2_8_36_1","doi-asserted-by":"crossref","first-page":"492","DOI":"10.1093\/aje\/kwm324","article-title":"Immortal time bias in pharmaco\u2010epidemiology","volume":"167","author":"Suissa S.","year":"2008","journal-title":"Am. J. Epidemiol."},{"key":"e_1_2_8_37_1","first-page":"809","article-title":"Joint modeling of longitudinal and time\u2010to\u2010event data: an OVERVIEW","volume":"14","author":"Tsiatis A.A.","year":"2004","journal-title":"Stat. Sin."},{"key":"e_1_2_8_38_1","doi-asserted-by":"crossref","first-page":"3437","DOI":"10.1002\/sim.3701","article-title":"Flexible modeling of the cumulative effects of time\u2010dependent exposures on the hazard","volume":"28","author":"Sylvestre M.P.","year":"2009","journal-title":"Stat. Med."},{"key":"e_1_2_8_39_1","doi-asserted-by":"crossref","first-page":"2389","DOI":"10.1002\/sim.2712","article-title":"Tutorial in biostatistics: competing risks and multi\u2010state models","volume":"26","author":"Putter H.","year":"2007","journal-title":"Stat. Med."},{"key":"e_1_2_8_40_1","doi-asserted-by":"crossref","first-page":"417","DOI":"10.1093\/ije\/dyp334","article-title":"Illustrating bias due to conditioning on a collider","volume":"39","author":"Cole S.R.","year":"2010","journal-title":"Int. J. Epidemiol."},{"key":"e_1_2_8_41_1","doi-asserted-by":"crossref","first-page":"615","DOI":"10.1097\/01.ede.0000135174.63482.43","article-title":"A structural approach to selection bias","volume":"15","author":"Hern\u00e1n M.A.","year":"2004","journal-title":"Epidemiology"},{"key":"e_1_2_8_42_1","doi-asserted-by":"crossref","first-page":"5749","DOI":"10.1038\/s41467-020-19478-2","article-title":"Collider bias undermines our understanding of COVID\u201019 disease risk and severity","volume":"11","author":"Griffith G.J.","year":"2020","journal-title":"Nat. Commun."},{"key":"e_1_2_8_43_1","doi-asserted-by":"crossref","first-page":"1282","DOI":"10.1001\/jama.2022.1820","article-title":"Collider Bias","volume":"327","author":"Holmberg M.J.","year":"2022","journal-title":"JAMA"},{"key":"e_1_2_8_44_1","doi-asserted-by":"crossref","first-page":"168","DOI":"10.1001\/jama.2020.11301","article-title":"Association of Angiotensin\u2010Converting Enzyme Inhibitor or angiotensin receptor blocker use with COVID\u201019 diagnosis and mortality","volume":"324","author":"Fosb\u00f8l E.L.","year":"2020","journal-title":"JAMA"},{"key":"e_1_2_8_45_1","doi-asserted-by":"crossref","DOI":"10.1017\/CBO9780511803161","volume-title":"Causality","author":"Pearl J.","year":"2009"},{"key":"e_1_2_8_46_1","doi-asserted-by":"crossref","first-page":"971","DOI":"10.1093\/jamia\/ocad013","article-title":"Healthcare utilization is a collider: an introduction to collider bias in EHR data reuse","volume":"30","author":"Weiskopf N.G.","year":"2023","journal-title":"J. Am. Med. Inform. Assoc."},{"key":"e_1_2_8_47_1","doi-asserted-by":"crossref","first-page":"928","DOI":"10.1161\/CIRCULATIONAHA.106.672402","article-title":"Use and misuse of the receiver operating characteristic curve in risk prediction","volume":"115","author":"Cook N.R.","year":"2007","journal-title":"Circulation"},{"key":"e_1_2_8_48_1","doi-asserted-by":"crossref","first-page":"1105","DOI":"10.1002\/sim.4154","article-title":"On the C\u2010statistics for evaluating overall adequacy of risk prediction procedures with censored survival data","volume":"30","author":"Uno H.","year":"2011","journal-title":"Stat. Med."},{"key":"e_1_2_8_49_1","first-page":"4768","volume-title":"Proceedings of the 31st International Conference on Neural Information Processing Systems","author":"Lundberg S.M.","year":"2017"},{"key":"e_1_2_8_50_1","doi-asserted-by":"crossref","first-page":"e745","DOI":"10.1016\/S2589-7500(21)00208-9","article-title":"The false hope of current approaches to explainable artificial intelligence in health care","volume":"3","author":"Ghassemi M.","year":"2021","journal-title":"Lancet Digit. Health"},{"key":"e_1_2_8_51_1","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pone.0118432","article-title":"The precision\u2010recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets","volume":"10","author":"Saito T.","year":"2015","journal-title":"PLoS One"},{"key":"e_1_2_8_52_1","first-page":"875","volume-title":"Data Mining and Knowledge Discovery Handbook","author":"Chawla N.V.","year":"2010"},{"key":"e_1_2_8_53_1","doi-asserted-by":"crossref","DOI":"10.1093\/jamiaopen\/ooaf116","article-title":"Predicting falls using electronic health records: a time series approach","volume":"8","author":"Hoover P.J.","year":"2025","journal-title":"JAMIA Open."},{"key":"e_1_2_8_54_1","doi-asserted-by":"crossref","DOI":"10.1136\/bmj.c221","article-title":"Meta\u2010analysis of individual participant data: rationale, conduct, and reporting","volume":"340","author":"Riley R.D.","year":"2010","journal-title":"BMJ"},{"key":"e_1_2_8_55_1","doi-asserted-by":"crossref","first-page":"1302","DOI":"10.1016\/j.jtho.2018.05.013","article-title":"PD\u2010L1 immunohistochemistry comparability study in real\u2010life clinical samples: results of blueprint phase 2 project","volume":"13","author":"Tsao M.S.","year":"2018","journal-title":"J. Thorac. Oncol."}],"container-title":["Clinical Pharmacology &amp; Therapeutics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/ascpt.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/cpt.70469","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ascpt.onlinelibrary.wiley.com\/doi\/full-xml\/10.1002\/cpt.70469","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ascpt.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/cpt.70469","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,3]],"date-time":"2026-09-03T12:24:43Z","timestamp":1788438283000},"score":1,"resource":{"primary":{"URL":"https:\/\/ascpt.onlinelibrary.wiley.com\/doi\/10.1002\/cpt.70469"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,3]]},"references-count":55,"alternative-id":["10.1002\/cpt.70469"],"URL":"https:\/\/doi.org\/10.1002\/cpt.70469","archive":["Portico"],"relation":{},"ISSN":["0009-9236","1532-6535"],"issn-type":[{"value":"0009-9236","type":"print"},{"value":"1532-6535","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,3]]},"assertion":[{"value":"2026-04-20","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-08-18","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-09-03","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"cpt.70469"}}