{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,7]],"date-time":"2026-11-07T04:01:27Z","timestamp":1794024087924,"version":"4.8.0"},"reference-count":40,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2024,8,2]],"date-time":"2024-08-02T00:00:00Z","timestamp":1722556800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2024,8,2]],"date-time":"2024-08-02T00:00:00Z","timestamp":1722556800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"DOI":"10.13039\/501100000038","name":"Natural Sciences and Engineering Research Council of Canada","doi-asserted-by":"publisher","award":["GR005815"],"award-info":[{"award-number":["GR005815"]}],"id":[{"id":"10.13039\/501100000038","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["BMC Med Res Methodol"],"abstract":"<jats:title>Abstract<\/jats:title><jats:sec>\n                <jats:title>Purpose<\/jats:title>\n                <jats:p>Propensity score matching is vital in epidemiological studies using observational data, yet its estimates relies on correct model-specification. This study assesses supervised deep learning models and unsupervised autoencoders for propensity score estimation, comparing them with traditional methods for bias and variance accuracy in treatment effect estimations.<\/jats:p>\n              <\/jats:sec><jats:sec>\n                <jats:title>Methods<\/jats:title>\n                <jats:p>Utilizing a plasmode simulation based on the Right Heart Catheterization dataset, under a variety of settings, we evaluated (1) a supervised deep learning architecture and (2) an unsupervised autoencoder, alongside two traditional methods: logistic regression and a spline-based method in estimating propensity scores for matching. Performance metrics included bias, standard errors, and coverage probability. The analysis was also extended to real-world data, with estimates compared to those obtained via a double robust approach.<\/jats:p>\n              <\/jats:sec><jats:sec>\n                <jats:title>Results<\/jats:title>\n                <jats:p>The analysis revealed that supervised deep learning models outperformed unsupervised autoencoders in variance estimation while maintaining comparable levels of bias. These results were supported by analyses of real-world data, where the supervised model\u2019s estimates closely matched those derived from conventional methods. Additionally, deep learning models performed well compared to traditional methods in settings where exposure was rare.<\/jats:p>\n              <\/jats:sec><jats:sec>\n                <jats:title>Conclusion<\/jats:title>\n                <jats:p>Supervised deep learning models hold promise in refining propensity score estimations in epidemiological research, offering nuanced confounder adjustment, especially in complex datasets. We endorse integrating supervised deep learning into epidemiological research and share reproducible codes for widespread use and methodological transparency.<\/jats:p>\n              <\/jats:sec>","DOI":"10.1186\/s12874-024-02284-5","type":"journal-article","created":{"date-parts":[[2024,8,2]],"date-time":"2024-08-02T08:36:00Z","timestamp":1722587760000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":11,"title":["Can supervised deep learning architecture outperform autoencoders in building propensity score models for matching?"],"prefix":"10.1186","volume":"24","author":[{"given":"Mohammad Ehsanul","family":"Karim","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2024,8,2]]},"reference":[{"issue":"10","key":"2284_CR1","doi-asserted-by":"publisher","first-page":"1685","DOI":"10.1002\/sim.6058","volume":"33","author":"JM Franklin","year":"2014","unstructured":"Franklin JM, Rassen JA, Ackermann D, Bartels DB, Schneeweiss S. Metrics for covariate balance in cohort studies of causal effects. Stat Med. 2014;33(10):1685\u201399.","journal-title":"Stat Med."},{"issue":"1","key":"2284_CR2","doi-asserted-by":"publisher","first-page":"7","DOI":"10.1177\/0962280210387717","volume":"21","author":"S Vansteelandt","year":"2012","unstructured":"Vansteelandt S, Bekaert M, Claeskens G. On model selection and model misspecification in causal inference. Stat Methods Med Res. 2012;21(1):7\u201330.","journal-title":"Stat Methods Med Res."},{"issue":"2","key":"2284_CR3","doi-asserted-by":"publisher","first-page":"108","DOI":"10.1093\/aje\/kwu253","volume":"181","author":"R Pirracchio","year":"2015","unstructured":"Pirracchio R, Petersen ML, Van Der Laan M. Improving propensity score estimators\u2019 robustness to model misspecification using super learner. Am J Epidemiol. 2015;181(2):108\u201319.","journal-title":"Am J Epidemiol."},{"issue":"4","key":"2284_CR4","first-page":"523","volume":"22","author":"JD Kang","year":"2007","unstructured":"Kang JD, Schafer JL. Demystifying double robustness: a comparison of alternative strategies for estimating a population mean from incomplete data. Stat Sci. 2007;22(4):523\u201339.","journal-title":"Stat Sci."},{"issue":"3","key":"2284_CR5","doi-asserted-by":"publisher","first-page":"337","DOI":"10.1002\/sim.3782","volume":"29","author":"BK Lee","year":"2010","unstructured":"Lee BK, Lessler J, Stuart EA. Improving propensity score weighting using machine learning. Stat Med. 2010;29(3):337\u201346.","journal-title":"Stat Med."},{"issue":"6","key":"2284_CR6","doi-asserted-by":"publisher","first-page":"1273","DOI":"10.1111\/1475-6773.13212","volume":"54","author":"KJ McConnell","year":"2019","unstructured":"McConnell KJ, Lindner S. Estimating treatment effects with machine learning. Health Serv Res. 2019;54(6):1273\u201382.","journal-title":"Health Serv Res."},{"key":"2284_CR7","doi-asserted-by":"publisher","first-page":"97","DOI":"10.1016\/j.jclinepi.2016.05.017","volume":"80","author":"T Schuster","year":"2016","unstructured":"Schuster T, Lowe WK, Platt RW. Propensity score model overfitting led to inflated variance of estimated odds ratios. J Clin Epidemiol. 2016;80:97\u2013106.","journal-title":"J Clin Epidemiol."},{"issue":"2","key":"2284_CR8","doi-asserted-by":"publisher","first-page":"191","DOI":"10.1097\/EDE.0000000000000787","volume":"29","author":"ME Karim","year":"2018","unstructured":"Karim ME, Pang M, Platt RW. Can we train machine learning methods to outperform the high-dimensional propensity score algorithm? Epidemiology. 2018;29(2):191\u20138.","journal-title":"Epidemiology."},{"issue":"4","key":"2284_CR9","doi-asserted-by":"publisher","first-page":"403","DOI":"10.1037\/1082-989X.9.4.403","volume":"9","author":"DF McCaffrey","year":"2004","unstructured":"McCaffrey DF, Ridgeway G, Morral AR. Propensity score estimation with boosted regression for evaluating causal effects in observational studies. Psychol Methods. 2004;9(4):403.","journal-title":"Psychol Methods."},{"issue":"5","key":"2284_CR10","doi-asserted-by":"publisher","first-page":"1798","DOI":"10.1111\/1475-6773.12068","volume":"48","author":"S Watkins","year":"2013","unstructured":"Watkins S, Jonsson-Funk M, Brookhart MA, Rosenberg SA, O\u2019Shea TM, Daniels J. An empirical comparison of tree-based methods for propensity score estimation. Health Serv Res. 2013;48(5):1798\u2013817.","journal-title":"Health Serv Res."},{"key":"2284_CR11","doi-asserted-by":"publisher","first-page":"69","DOI":"10.1007\/s10742-014-0124-y","volume":"14","author":"Y Zhu","year":"2014","unstructured":"Zhu Y, Ghosh D, Mitra N, Mukherjee B. A data-adaptive strategy for inverse weighted estimation of causal effects. Health Serv Outcome Res Methodol. 2014;14:69\u201391.","journal-title":"Health Serv Outcome Res Methodol."},{"issue":"1","key":"2284_CR12","doi-asserted-by":"publisher","first-page":"164","DOI":"10.1080\/00273171.2013.752263","volume":"48","author":"BS Keller","year":"2013","unstructured":"Keller BS, Kim JS, Steiner PM. Data mining alternatives to logistic regression for propensity score estimation: Neural networks and support vector machines. Multivar Behav Res. 2013;48(1):164.","journal-title":"Multivar Behav Res."},{"key":"2284_CR13","doi-asserted-by":"publisher","unstructured":"Chen K, Yin Q, Long Q. Covariate-balancing-aware interpretable deep learning models for treatment effect estimation. Stat Biosci. 2023:1\u201319. https:\/\/doi.org\/10.1007\/s12561-023-09394-6.\u00a0Accessed 1 Jan 2024.","DOI":"10.1007\/s12561-023-09394-6"},{"key":"2284_CR14","doi-asserted-by":"publisher","first-page":"2813","DOI":"10.1109\/ACCESS.2021.3140189","volume":"10","author":"A Whata","year":"2022","unstructured":"Whata A, Chimedza C. Evaluating uses of deep learning methods for causal inference. IEEE Access. 2022;10:2813\u201327.","journal-title":"IEEE Access."},{"issue":"1","key":"2284_CR15","doi-asserted-by":"publisher","first-page":"164","DOI":"10.1080\/00273171.2021.2011695","volume":"57","author":"A Guzman-Alvarez","year":"2022","unstructured":"Guzman-Alvarez A, Qin X, Scott PW. Deep Neural Networks for Propensity Score Estimation. Multivar Behav Res. 2022;57(1):164\u20135.","journal-title":"Multivar Behav Res."},{"issue":"6","key":"2284_CR16","doi-asserted-by":"publisher","first-page":"1197","DOI":"10.1093\/jamia\/ocaa346","volume":"28","author":"S Ghosh","year":"2021","unstructured":"Ghosh S, Bian J, Guo Y, Prosperi M. Deep propensity network using a sparse autoencoder for estimation of treatment effects. J Am Med Inform Assoc. 2021;28(6):1197\u2013206.","journal-title":"J Am Med Inform Assoc."},{"issue":"3","key":"2284_CR17","first-page":"378","volume":"32","author":"J Weberpals","year":"2021","unstructured":"Weberpals J, Becker T, Davies J, Schmich F, R\u00fcttinger D, Theis FJ, et al. Deep Learning-based Propensity Scores for Confounding Control in Comparative Effectiveness Research: A Large-scale. Real-world Data Study Epidemiology. 2021;32(3):378\u201388.","journal-title":"Real-world Data Study. Epidemiology."},{"key":"2284_CR18","doi-asserted-by":"publisher","first-page":"100020","DOI":"10.1016\/j.cmpbup.2021.100020","volume":"1","author":"S Ghosh","year":"2021","unstructured":"Ghosh S, Boucher C, Bian J, Prosperi M. Propensity score synthetic augmentation matching using generative adversarial networks (PSSAM-GAN). Comput Methods Prog Biomed Updat. 2021;1:100020.","journal-title":"Comput Methods Prog Biomed Updat."},{"key":"2284_CR19","unstructured":"Ramachandra V. Deep learning for causal inference. 2018. arXiv preprint arXiv:1803.00149."},{"key":"2284_CR20","doi-asserted-by":"publisher","first-page":"100048","DOI":"10.1016\/j.ostima.2022.100048","volume":"2","author":"B Mohajer","year":"2022","unstructured":"Mohajer B, Dolatashahi M, Moradi K, Najafzadeh N, Enj J, Zikria B, et al. Thigh muscle changes can worsen subsequent knee oa clinical outcomes: use of deep learning and propensity-score matching on OAI data. Osteoarthr Imaging. 2022;2:100048.","journal-title":"Osteoarthr Imaging."},{"key":"2284_CR21","doi-asserted-by":"publisher","first-page":"219","DOI":"10.1016\/j.csda.2013.10.018","volume":"72","author":"JM Franklin","year":"2014","unstructured":"Franklin JM, Schneeweiss S, Polinski JM, Rassen JA. Plasmode simulation for the evaluation of pharmacoepidemiologic methods in complex healthcare databases. Comput Stat Data Anal. 2014;72:219\u201326.","journal-title":"Comput Stat Data Anal."},{"issue":"11","key":"2284_CR22","doi-asserted-by":"publisher","first-page":"889","DOI":"10.1001\/jama.1996.03540110043030","volume":"276","author":"AF Connors","year":"1996","unstructured":"Connors AF, Speroff T, Dawson NV, Thomas C, Harrell FE, Wagner D, et al. The effectiveness of right heart catheterization in the initial care of critically III patients. JAMA. 1996;276(11):889\u201397.","journal-title":"JAMA."},{"issue":"11","key":"2284_CR23","doi-asserted-by":"publisher","first-page":"2074","DOI":"10.1002\/sim.8086","volume":"38","author":"TP Morris","year":"2019","unstructured":"Morris TP, White IR, Crowther MJ. Using simulation studies to evaluate statistical methods. Stat Med. 2019;38(11):2074\u2013102.","journal-title":"Stat Med."},{"key":"2284_CR24","doi-asserted-by":"crossref","unstructured":"White IR, Pham TM, Quartagno M, Morris TP. How to check a simulation study. Int J Epidemiol. 2023:dyad134.","DOI":"10.31219\/osf.io\/cbr72"},{"issue":"2","key":"2284_CR25","doi-asserted-by":"publisher","first-page":"150","DOI":"10.1002\/pst.433","volume":"10","author":"PC Austin","year":"2011","unstructured":"Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat. 2011;10(2):150\u201361.","journal-title":"Pharm Stat."},{"issue":"8","key":"2284_CR26","doi-asserted-by":"publisher","first-page":"S84","DOI":"10.1016\/j.jclinepi.2013.01.013","volume":"66","author":"EA Stuart","year":"2013","unstructured":"Stuart EA, Lee BK, Leacy FP. Prognostic score-based balance measures can be a useful diagnostic for propensity score methods in comparative effectiveness research. J Clin Epidemiol. 2013;66(8):S84\u201390.","journal-title":"J Clin Epidemiol."},{"issue":"1","key":"2284_CR27","doi-asserted-by":"publisher","first-page":"201","DOI":"10.1177\/0962280214543508","volume":"26","author":"PC Austin","year":"2017","unstructured":"Austin PC. Double propensity-score adjustment: a solution to design bias or bias due to incomplete matching. Stat Methods Med Res. 2017;26(1):201\u201322.","journal-title":"Stat Methods Med Res."},{"key":"2284_CR28","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s12874-017-0338-0","volume":"17","author":"TL Nguyen","year":"2017","unstructured":"Nguyen TL, Collins GS, Spence J, Daur\u00e8s JP, Devereaux P, Landais P, et al. Double-adjustment in propensity score matching analysis: choosing a threshold for considering residual imbalance. BMC Med Res Methodol. 2017;17:1\u20138.","journal-title":"BMC Med Res Methodol."},{"issue":"9","key":"2284_CR29","doi-asserted-by":"publisher","first-page":"1317","DOI":"10.1177\/1352458520972557","volume":"28","author":"ME Karim","year":"2022","unstructured":"Karim ME, Pellegrini F, Platt RW, Simoneau G, Rouette J, de Moor C. The use and quality of reporting of propensity score methods in multiple sclerosis literature: a review. Mult Scler J. 2022;28(9):1317\u201323.","journal-title":"Mult Scler J."},{"issue":"1","key":"2284_CR30","first-page":"1","volume":"19","author":"JH Friedman","year":"1991","unstructured":"Friedman JH. Multivariate adaptive regression splines. Ann Stat. 1991;19(1):1\u201367.","journal-title":"Ann Stat."},{"key":"2284_CR31","unstructured":"Milborrow S. Earth: multivariate adaptive regression splines. 2023. R package version 5.3.2.\u00a0https:\/\/CRAN.R-project.org\/package=earth. Accessed 1 Jan 2024."},{"issue":"1","key":"2284_CR32","doi-asserted-by":"publisher","first-page":"65","DOI":"10.1093\/aje\/kww165","volume":"185","author":"MS Schuler","year":"2017","unstructured":"Schuler MS, Rose S. Targeted maximum likelihood estimation for causal inference in observational studies. Am J Epidemiol. 2017;185(1):65\u201373.","journal-title":"Am J Epidemiol."},{"issue":"1","key":"2284_CR33","first-page":"8","volume":"5","author":"HA Frank","year":"2024","unstructured":"Frank HA, Karim ME. Implementing TMLE in the presence of a continuous outcome. Res Methods Med Health Sci. 2024;5(1):8\u201319.","journal-title":"Res Methods Med Health Sci."},{"issue":"4","key":"2284_CR34","doi-asserted-by":"publisher","first-page":"1276","DOI":"10.1093\/ije\/dyad023","volume":"52","author":"RV Phillips","year":"2023","unstructured":"Phillips RV, van der Laan MJ, Lee H, Gruber S. Practical considerations for specifying a super learner. Int J Epidemiol. 2023;52(4):1276\u201385.","journal-title":"Int J Epidemiol."},{"key":"2284_CR35","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s12874-016-0135-1","volume":"16","author":"D Hajage","year":"2016","unstructured":"Hajage D, Tubach F, Steg PG, Bhatt DL, De Rycke Y. On the use of propensity scores in case of rare exposure. BMC Med Res Methodol. 2016;16:1\u201316.","journal-title":"BMC Med Res Methodol."},{"issue":"2","key":"2284_CR36","doi-asserted-by":"publisher","first-page":"246","DOI":"10.1016\/j.ijcard.2007.05.032","volume":"125","author":"A Ahmed","year":"2008","unstructured":"Ahmed A, Young JB, Love TE, Levesque R, Pitt B. A propensity-matched study of the effects of chronic diuretic therapy on mortality and hospitalization in older adults with heart failure. Int J Cardiol. 2008;125(2):246\u201353.","journal-title":"Int J Cardiol."},{"issue":"4","key":"2284_CR37","doi-asserted-by":"publisher","first-page":"435","DOI":"10.1017\/pan.2019.11","volume":"27","author":"G King","year":"2019","unstructured":"King G, Nielsen R. Why propensity scores should not be used for matching. Polit Anal. 2019;27(4):435\u201354.","journal-title":"Polit Anal."},{"issue":"6","key":"2284_CR38","doi-asserted-by":"publisher","first-page":"1057","DOI":"10.1002\/sim.6004","volume":"33","author":"PC Austin","year":"2014","unstructured":"Austin PC. A comparison of 12 algorithms for matching on the propensity score. Stat Med. 2014;33(6):1057\u201369.","journal-title":"Stat Med."},{"issue":"1","key":"2284_CR39","doi-asserted-by":"publisher","first-page":"6654","DOI":"10.1038\/s41467-021-26990-6","volume":"12","author":"L Lu","year":"2021","unstructured":"Lu L, Dercle L, Zhao B, Schwartz LH. Deep learning for the prediction of early on-treatment response in metastatic colorectal cancer from serial medical imaging. Nat Commun. 2021;12(1):6654.","journal-title":"Nat Commun."},{"issue":"1","key":"2284_CR40","doi-asserted-by":"publisher","first-page":"15031","DOI":"10.1038\/s41598-023-41544-0","volume":"13","author":"B Son","year":"2023","unstructured":"Son B, Myung J, Shin Y, Kim S, Kim SH, Chung JM, et al. Improved patient mortality predictions in emergency departments with deep learning data-synthesis and ensemble models. Sci Rep. 2023;13(1):15031.","journal-title":"Sci Rep."}],"container-title":["BMC Medical Research Methodology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s12874-024-02284-5.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1186\/s12874-024-02284-5\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s12874-024-02284-5.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,8,2]],"date-time":"2024-08-02T08:38:19Z","timestamp":1722587899000},"score":1,"resource":{"primary":{"URL":"https:\/\/bmcmedresmethodol.biomedcentral.com\/articles\/10.1186\/s12874-024-02284-5"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,8,2]]},"references-count":40,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2024,12]]}},"alternative-id":["2284"],"URL":"https:\/\/doi.org\/10.1186\/s12874-024-02284-5","relation":{},"ISSN":["1471-2288"],"issn-type":[{"value":"1471-2288","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,8,2]]},"assertion":[{"value":"5 March 2024","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"15 July 2024","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"2 August 2024","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The analysis conducted on secondary and de-identified data is exempt from research ethics approval requirements. Ethics for this study was covered by item 7.10.3 in University of British Columbia\u2019s Policy #89: Research and Other Studies Involving Human Subjects 19 and Article 2.2 in of the Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans (TCPS2).","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics approval and consent to participate"}},{"value":"Given the secondary (from SUPPORT study) and anonymized nature of the data, individual consent for publication is not applicable. All results are presented in aggregate form, and no individual data are disclosed in this publication.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent for publication"}},{"value":"Over the past three years, MEK has received consulting fees from Biogen Inc. for consulting unrelated to this current work.","order":4,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"167"}}