{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,3]],"date-time":"2026-04-03T21:45:11Z","timestamp":1775252711192,"version":"3.50.1"},"reference-count":29,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2022,12,22]],"date-time":"2022-12-22T00:00:00Z","timestamp":1671667200000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2022,12,22]],"date-time":"2022-12-22T00:00:00Z","timestamp":1671667200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["npj Digit. Med."],"abstract":"<jats:title>Abstract<\/jats:title><jats:p>Deep learning has been shown to accurately assess \u201chidden\u201d phenotypes from medical imaging beyond traditional clinician interpretation. Using large echocardiography datasets from two healthcare systems, we test whether it is possible to predict age, race, and sex from cardiac ultrasound images using deep learning algorithms and assess the impact of varying confounding variables. Using a total of 433,469 videos from Cedars-Sinai Medical Center and 99,909 videos from Stanford Medical Center, we trained video-based convolutional neural networks to predict age, sex, and race. We found that deep learning models were able to identify age and sex, while unable to reliably predict race. Without considering confounding differences between categories, the AI model predicted sex with an AUC of 0.85 (95% CI 0.84\u20130.86), age with a mean absolute error of 9.12 years (95% CI 9.00\u20139.25), and race with AUCs ranging from 0.63 to 0.71. When predicting race, we show that tuning the proportion of confounding variables (age or sex) in the training data significantly impacts model AUC (ranging from 0.53 to 0.85), while sex and age prediction was not particularly impacted by adjusting race proportion in the training dataset AUC of 0.81\u20130.83 and 0.80\u20130.84, respectively. This suggests significant proportion of AI\u2019s performance on predicting race could come from confounding features being detected. Further work remains to identify the particular imaging features that associate with demographic information and to better understand the risks of demographic identification in medical AI as it pertains to potentially perpetuating bias and disparities.<\/jats:p>","DOI":"10.1038\/s41746-022-00720-8","type":"journal-article","created":{"date-parts":[[2022,12,22]],"date-time":"2022-12-22T15:02:51Z","timestamp":1671721371000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":45,"title":["Confounders mediate AI prediction of demographics in medical imaging"],"prefix":"10.1038","volume":"5","author":[{"given":"Grant","family":"Duffy","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6592-1172","authenticated-orcid":false,"given":"Shoa L.","family":"Clarke","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Matthew","family":"Christensen","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6150-761X","authenticated-orcid":false,"given":"Bryan","family":"He","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5782-7437","authenticated-orcid":false,"given":"Neal","family":"Yuan","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4977-036X","authenticated-orcid":false,"given":"Susan","family":"Cheng","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3813-7518","authenticated-orcid":false,"given":"David","family":"Ouyang","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"297","published-online":{"date-parts":[[2022,12,22]]},"reference":[{"key":"720_CR1","doi-asserted-by":"publisher","first-page":"31","DOI":"10.1038\/s41591-021-01614-0","volume":"28","author":"P Rajpurkar","year":"2022","unstructured":"Rajpurkar, P., Chen, E., Banerjee, O. & Topol, E. J. AI in health and medicine. Nat. Med. 28, 31\u201338 (2022).","journal-title":"Nat. Med."},{"key":"720_CR2","doi-asserted-by":"publisher","first-page":"211","DOI":"10.1007\/s11263-015-0816-y","volume":"115","author":"O Russakovsky","year":"2015","unstructured":"Russakovsky, O. et al. ImageNet large scale visual recognition challenge. Int. J. Comput. Vis. 115, 211\u2013252 (2015).","journal-title":"Int. J. Comput. Vis."},{"key":"720_CR3","doi-asserted-by":"crossref","unstructured":"Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J. & Wojna, Z. Rethinking the inception architecture for computer vision. Preprint at arXiv https:\/\/arxiv.org\/abs\/1512.00567 (2015).","DOI":"10.1109\/CVPR.2016.308"},{"key":"720_CR4","doi-asserted-by":"crossref","unstructured":"He, K., Zhang, X., Ren, S. & Sun, J. Deep residual learning for image recognition. Preprint at arXiv https:\/\/arxiv.org\/abs\/1512.03385 (2015).","DOI":"10.1109\/CVPR.2016.90"},{"key":"720_CR5","unstructured":"Lin, T.-Y. et al. Microsoft COCO: common objects in context. Preprint at arXiv https:\/\/arxiv.org\/abs\/1405.0312 (2014)."},{"key":"720_CR6","doi-asserted-by":"publisher","first-page":"158","DOI":"10.1038\/s41551-018-0195-0","volume":"2","author":"R Poplin","year":"2018","unstructured":"Poplin, R. et al. Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning. Nat. Biomed. Eng. 2, 158\u2013164 (2018).","journal-title":"Nat. Biomed. Eng."},{"key":"720_CR7","doi-asserted-by":"crossref","unstructured":"Weston Hughes, J. et al. Deep learning prediction of biomarkers from echocardiogram videos. EBioMedicine 73, 103613 (2021).","DOI":"10.1016\/j.ebiom.2021.103613"},{"key":"720_CR8","doi-asserted-by":"publisher","first-page":"10","DOI":"10.1038\/s41746-019-0216-8","volume":"3","author":"A Ghorbani","year":"2020","unstructured":"Ghorbani, A. et al. Deep learning interpretation of echocardiograms. NPJ Digit Med 3, 10 (2020).","journal-title":"NPJ Digit Med"},{"key":"720_CR9","doi-asserted-by":"publisher","unstructured":"Yi, P. H. et al. Radiology \u201cforensics\u201d: determination of age and sex from chest radiographs using deep learning. Emerg. Radiol. https:\/\/doi.org\/10.1007\/s10140-021-01953-y (2021).","DOI":"10.1007\/s10140-021-01953-y"},{"key":"720_CR10","doi-asserted-by":"publisher","first-page":"2226","DOI":"10.1016\/j.jcmg.2021.01.008","volume":"14","author":"VK Raghu","year":"2021","unstructured":"Raghu, V. K., Weiss, J., Hoffmann, U., Aerts, H. J. W. L. & Lu, M. T. Deep learning to estimate biological age from chest radiographs. JACC Cardiovasc. Imaging 14, 2226\u20132236 (2021).","journal-title":"JACC Cardiovasc. Imaging"},{"key":"720_CR11","doi-asserted-by":"publisher","first-page":"e007284","DOI":"10.1161\/CIRCEP.119.007284","volume":"12","author":"ZI Attia","year":"2019","unstructured":"Attia, Z. I. et al. Age and sex estimation using artificial intelligence from standard 12-lead ECGs. Circ. Arrhythm. Electrophysiol. 12, e007284 (2019).","journal-title":"Circ. Arrhythm. Electrophysiol."},{"key":"720_CR12","unstructured":"Banerjee, I. et al. Reading race: AI recognises patient\u2019s racial identity in medical images. Preprint at arXiv https:\/\/www.thelancet.com\/journals\/landig\/article\/PIIS2589-7500(22)00063-2\/fulltext. 10356 (2021)."},{"key":"720_CR13","doi-asserted-by":"publisher","first-page":"324","DOI":"10.1038\/d41586-018-05707-8","volume":"559","author":"J Zou","year":"2018","unstructured":"Zou, J. & Schiebinger, L. AI can be sexist and racist \u2014 it\u2019s time to make it fair. Nature 559, 324\u2013326 (2018).","journal-title":"Nature"},{"key":"720_CR14","doi-asserted-by":"crossref","unstructured":"Seyyed-Kalantari, L., Liu, G., McDermott, M., Chen, I. Y. & Ghassemi, M. CheXclusion: fairness gaps in deep chest X-ray classifiers. Preprint at arXiv https:\/\/arxiv.org\/abs\/2003.00827 (2020).","DOI":"10.1142\/9789811232701_0022"},{"key":"720_CR15","doi-asserted-by":"publisher","first-page":"136","DOI":"10.1038\/s41591-020-01192-7","volume":"27","author":"E Pierson","year":"2021","unstructured":"Pierson, E., Cutler, D. M., Leskovec, J., Mullainathan, S. & Obermeyer, Z. An algorithmic approach to reducing unexplained pain disparities in underserved populations. Nat. Med. 27, 136\u2013140 (2021).","journal-title":"Nat. Med."},{"key":"720_CR16","doi-asserted-by":"publisher","first-page":"582","DOI":"10.1038\/s41591-021-01312-x","volume":"27","author":"E Wu","year":"2021","unstructured":"Wu, E. et al. How medical AI devices are evaluated: limitations and recommendations from an analysis of FDA approvals. Nat. Med. 27, 582\u2013584 (2021).","journal-title":"Nat. Med."},{"key":"720_CR17","doi-asserted-by":"publisher","first-page":"447","DOI":"10.1126\/science.aax2342","volume":"366","author":"Z Obermeyer","year":"2019","unstructured":"Obermeyer, Z., Powers, B., Vogeli, C. & Mullainathan, S. Dissecting racial bias in an algorithm used to manage the health of populations. Science 366, 447\u2013453 (2019).","journal-title":"Science"},{"key":"720_CR18","doi-asserted-by":"publisher","first-page":"252","DOI":"10.1038\/s41586-020-2145-8","volume":"580","author":"D Ouyang","year":"2020","unstructured":"Ouyang, D. et al. Video-based AI for beat-to-beat assessment of cardiac function. Nature 580, 252\u2013256 (2020).","journal-title":"Nature"},{"key":"720_CR19","doi-asserted-by":"crossref","unstructured":"Duffy, G. et al. High-throughput precision phenotyping of left ventricular hypertrophy with cardiovascular deep learning. JAMA Cardiol. 7, 386\u2013395 (2022).","DOI":"10.1001\/jamacardio.2021.6059"},{"key":"720_CR20","doi-asserted-by":"publisher","first-page":"191","DOI":"10.1161\/CIRCIMAGING.108.819938","volume":"2","author":"S Cheng","year":"2009","unstructured":"Cheng, S. et al. Age-related left ventricular remodeling and associated risk for cardiovascular outcomes: the Multi-Ethnic Study of Atherosclerosis. Circ. Cardiovasc. Imaging 2, 191\u2013198 (2009).","journal-title":"Circ. Cardiovasc. Imaging"},{"key":"720_CR21","doi-asserted-by":"publisher","first-page":"1396.e2","DOI":"10.1016\/j.echo.2019.08.012","volume":"32","author":"FM Asch","year":"2019","unstructured":"Asch, F. M. et al. Similarities and differences in left ventricular size and function among races and nationalities: results of the World Alliance Societies of Echocardiography Normal Values Study. J. Am. Soc. Echocardiogr. 32, 1396.e2\u20131406.e2 (2019).","journal-title":"J. Am. Soc. Echocardiogr."},{"key":"720_CR22","doi-asserted-by":"publisher","DOI":"10.1186\/1471-2342-9-2","volume":"9","author":"PA Cain","year":"2009","unstructured":"Cain, P. A. et al. Age and gender specific normal values of left ventricular mass, volume and function for gradient echo magnetic resonance imaging: a cross sectional study. BMC Med. Imaging 9, 2 (2009).","journal-title":"BMC Med. Imaging"},{"key":"720_CR23","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s12968-020-00683-3","volume":"22","author":"N Kawel-Boehm","year":"2020","unstructured":"Kawel-Boehm, N. et al. Reference ranges (\u201cnormal values\u201d) for cardiovascular magnetic resonance (CMR) in adults and children: 2020 update. J. Cardiovasc. Magn. Reson. 22, 1\u201363 (2020).","journal-title":"J. Cardiovasc. Magn. Reson."},{"key":"720_CR24","doi-asserted-by":"publisher","first-page":"1223","DOI":"10.1016\/j.echo.2020.06.008","volume":"33","author":"T Miyoshi","year":"2020","unstructured":"Miyoshi, T. et al. Left ventricular diastolic function in healthy adult individuals: results of the World Alliance Societies of Echocardiography Normal Values Study. J. Am. Soc. Echocardiogr. 33, 1223\u20131233 (2020).","journal-title":"J. Am. Soc. Echocardiogr."},{"key":"720_CR25","doi-asserted-by":"publisher","first-page":"501","DOI":"10.1056\/NEJMp2116224","volume":"386","author":"A Deyrup","year":"2022","unstructured":"Deyrup, A. & Graves, J. L. Jr. Racial biology and medical misconceptions. N. Engl. J. Med. 386, 501\u2013503 (2022).","journal-title":"N. Engl. J. Med."},{"key":"720_CR26","doi-asserted-by":"publisher","unstructured":"Ganapathi, S. et al. Tackling bias in AI health datasets through the STANDING Together initiative. Nat. Med. https:\/\/doi.org\/10.1038\/s41591-022-01987-w (2022).","DOI":"10.1038\/s41591-022-01987-w"},{"key":"720_CR27","doi-asserted-by":"publisher","first-page":"874","DOI":"10.1056\/NEJMms2004740","volume":"383","author":"DA Vyas","year":"2020","unstructured":"Vyas, D. A., Eisenstein, L. G. & Jones, D. S. Hidden in plain sight - reconsidering the use of race correction in clinical algorithms. N. Engl. J. Med. 383, 874\u2013882 (2020).","journal-title":"N. Engl. J. Med."},{"key":"720_CR28","doi-asserted-by":"publisher","first-page":"768","DOI":"10.1056\/NEJMms2025396","volume":"384","author":"ZD Bailey","year":"2021","unstructured":"Bailey, Z. D., Feldman, J. M. & Bassett, M. T. How structural racism works - racist policies as a root cause of U.s. racial health inequities. N. Engl. J. Med. 384, 768\u2013773 (2021).","journal-title":"N. Engl. J. Med."},{"key":"720_CR29","unstructured":"Ouyang, D. ConvertDICOMToAVI.ipynb at master \u00b7 echonet\/dynamic. Github. https:\/\/github.com\/echonet\/dynamic\/blob\/master\/scripts\/ConvertDICOMToAVI.ipynb. Accessed Jan 1, 2022."}],"container-title":["npj Digital Medicine"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41746-022-00720-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41746-022-00720-8","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41746-022-00720-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,12,22]],"date-time":"2022-12-22T15:13:51Z","timestamp":1671722031000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41746-022-00720-8"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,12,22]]},"references-count":29,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2022,12]]}},"alternative-id":["720"],"URL":"https:\/\/doi.org\/10.1038\/s41746-022-00720-8","relation":{},"ISSN":["2398-6352"],"issn-type":[{"value":"2398-6352","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,12,22]]},"assertion":[{"value":"27 July 2022","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"2 November 2022","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"22 December 2022","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"The authors declare no competing interests.","order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"188"}}