{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,8]],"date-time":"2026-07-08T23:19:35Z","timestamp":1783552775342,"version":"3.55.0"},"reference-count":51,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T00:00:00Z","timestamp":1777852800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,7,9]],"date-time":"2026-07-09T00:00:00Z","timestamp":1783555200000},"content-version":"vor","delay-in-days":66,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Nat Commun"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Protecting patients\u2019 right to be forgotten (RtbF) in the digital era underscores algorithms to delete health records from deployed artificial intelligence (AI) models upon requests, a paradigm also known as machine unlearning. Such forgetfulness, however, will alter AI models\u2019 views of subpopulations and could yield discriminating clinical decisions that undermine healthcare equality and endanger patient trust, a common phenomenon confirmed by our thorough investigations. This observation reveals an ethical dilemma in clinical AI: protecting the RtbF of certain patients will compromise the delivery of equitable decisions and, thus, is against the AI fairness principle, necessitating a safeguarding approach. To this end, we propose a fair unlearning strategy to effectively remove medical records from trained models while mitigating decision biases to improve algorithmic equality. This strategy avoids negative interventions between potentially conflicting objectives by enforcing their gradient orthogonality. We perform extensive evaluations on real-world, multi-hospital datasets of rapid COVID-19 screening, in-hospital mortality and shock predictions. Compared with state-of-the-art approaches, our method can teach models to more effectively forget patient records with better predictive performance while, more importantly, mitigating demographic unfairness across subpopulations of ethnicities and sites (hospitals), accommodating the first generalised solution for the dilemma of RtbF and fairness.<\/jats:p>","DOI":"10.1038\/s41467-026-72601-7","type":"journal-article","created":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T16:35:57Z","timestamp":1777912557000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Mitigating algorithmic unfairness arising from forgetfulness of medical records in clinical artificial intelligence"],"prefix":"10.1038","volume":"17","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-4460-7907","authenticated-orcid":false,"given":"Yixuan","family":"Chen","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7006-1947","authenticated-orcid":false,"given":"Anshul","family":"Thakur","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2391-5361","authenticated-orcid":false,"given":"Andrew A. S.","family":"Soltan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yifei","family":"Shen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3103-8442","authenticated-orcid":false,"given":"Dongsheng","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8897-5931","authenticated-orcid":false,"given":"Mingzhi","family":"Dong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Li","family":"Shang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9848-8555","authenticated-orcid":false,"given":"David A.","family":"Clifton","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6220-029X","authenticated-orcid":false,"given":"Yujiang","family":"Wang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,5,4]]},"reference":[{"key":"72601_CR1","doi-asserted-by":"publisher","first-page":"3","DOI":"10.1080\/15710880600608230","volume":"2","author":"LJ Bannon","year":"2006","unstructured":"Bannon, L. J. Forgetting as a feature, not a bug: the dualityof memory and implications for ubiquitous computing. CoDesign 2, 3\u201315 (2006).","journal-title":"CoDesign"},{"key":"72601_CR2","doi-asserted-by":"crossref","unstructured":"Ricoeur, P. et al. Memory, History, Forgetting. (University of Chicago Press, 2004).","DOI":"10.7208\/chicago\/9780226713465.001.0001"},{"key":"72601_CR3","doi-asserted-by":"crossref","unstructured":"Mayer-Sch\u00f6nberger, V. et al. Delete: The Virtue of Forgetting in the Digital Age. (Princeton University Press, 2011).","DOI":"10.1515\/9781400838455"},{"key":"72601_CR4","doi-asserted-by":"crossref","unstructured":"Voigt, P. & VondemBussche, A. The EU General Data Protection Regulation (GDPR): A Practical Guide (2017).","DOI":"10.1007\/978-3-319-57959-7"},{"key":"72601_CR5","doi-asserted-by":"publisher","first-page":"tyy001","DOI":"10.1093\/cybsec\/tyy001","volume":"4","author":"E Politou","year":"2018","unstructured":"Politou, E., Alepis, E. & Patsakis, C. Forgetting personal data and revoking consent under the gdpr: Challenges and proposed solutions. J. Cybersecur. 4, tyy001 (2018).","journal-title":"J. Cybersecur."},{"key":"72601_CR6","doi-asserted-by":"crossref","unstructured":"Song, C., Ristenpart, T., and Shmatikov, V. Machine learning models that remember too much. In Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security. (2017).","DOI":"10.1145\/3133956.3134077"},{"key":"72601_CR7","unstructured":"Carlini, N., Liu, C., Erlingsson, \u00da., Kos, J. & Song, D. The secret sharer: Evaluating and testing unintended memorization in neural networks. In Proceedings of the 28th USENIX security symposium (USENIX security 19). (2019)."},{"key":"72601_CR8","doi-asserted-by":"publisher","first-page":"eadh8601","DOI":"10.1126\/sciadv.adh8601","volume":"10","author":"J Zhou","year":"2024","unstructured":"Zhou, J. Ppml-omics: a privacy-preserving federated machine learning method protects patients\u2019 privacy in omic data. Sci. Adv. 10, eadh8601 (2024).","journal-title":"Sci. Adv."},{"key":"72601_CR9","doi-asserted-by":"publisher","first-page":"1201","DOI":"10.1056\/NEJMra2302038","volume":"388","author":"CJ Haug","year":"2023","unstructured":"Haug, C. J. & Drazen, J. M. Artificial intelligence and machine learning in clinical medicine, 2023. N. Engl. J. Med. 388, 1201\u20131208 (2023).","journal-title":"N. Engl. J. Med."},{"key":"72601_CR10","doi-asserted-by":"publisher","first-page":"930","DOI":"10.1038\/s44222-023-00096-8","volume":"1","author":"AH Song","year":"2023","unstructured":"Song, A. H. Artificial intelligence for digital and computational pathology. Nat. Rev. Bioeng. 1, 930\u2013949 (2023).","journal-title":"Nat. Rev. Bioeng."},{"key":"72601_CR11","doi-asserted-by":"crossref","unstructured":"Eisemann, N. et al. Nationwide real-world implementation of ai for cancer detection in population-based mammography screening. Nat. Med. 31, 917\u2013924 (2025).","DOI":"10.1038\/s41591-024-03408-6"},{"key":"72601_CR12","unstructured":"Mayer-Sch\u00f6nberger, V. et al. Omission of search results is not a \u201cright to be forgotten\u201d or the end of google. The Guardian May 13 (2014). Available at: https:\/\/www.theguardian.com\/commentisfree\/2014\/may\/13\/omission-of-search-results-no-right-to-be-forgotten (accessed February 18, 2025)."},{"key":"72601_CR13","doi-asserted-by":"publisher","first-page":"337","DOI":"10.1007\/s42979-023-01767-4","volume":"4","author":"H Zhang","year":"2023","unstructured":"Zhang, H., Nakamura, T., Isohara, T. & Sakurai, K. A review on machine unlearning. SN Comput. Sci. 4, 337 (2023).","journal-title":"SN Comput. Sci."},{"key":"72601_CR14","doi-asserted-by":"crossref","unstructured":"Bourtoule, L. et al. Machine unlearning. In Proceedings 2021 IEEE symposium on security and privacy (SP) (2021).","DOI":"10.1109\/SP40001.2021.00019"},{"key":"72601_CR15","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-024-01270-x","volume":"7","author":"V Muralidharan","year":"2024","unstructured":"Muralidharan, V. A scoping review of reporting gaps in fda-approved ai medical devices. Npj Digit. Med. 7, 273 (2024).","journal-title":"Npj Digit. Med."},{"key":"72601_CR16","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01459-8","volume":"8","author":"FE Rademakers","year":"2025","unstructured":"Rademakers, F. E. Core-md clinical risk score for regulatory evaluation of artificial intelligence-based medical device software. Npj Digit. Med. 8, 90 (2025).","journal-title":"Npj Digit. Med."},{"key":"72601_CR17","doi-asserted-by":"publisher","first-page":"e78","DOI":"10.1016\/S2589-7500(20)30274-0","volume":"3","author":"AA Soltan","year":"2021","unstructured":"Soltan, A. A. Rapid triage for covid-19 using routine clinical data for patients attending hospital: development and prospective validation of an artificial intelligence screening test. Lancet Digit. Health 3, e78\u2013e87 (2021).","journal-title":"Lancet Digit. Health"},{"key":"72601_CR18","doi-asserted-by":"publisher","first-page":"e266","DOI":"10.1016\/S2589-7500(21)00272-7","volume":"4","author":"AA Soltan","year":"2022","unstructured":"Soltan, A. A. Real-world evaluation of rapid and laboratory-free covid-19 triage for emergency care: external validation and pilot deployment of artificial intelligence driven screening. Lancet Digit. Health 4, e266\u2013e278 (2022).","journal-title":"Lancet Digit. Health"},{"key":"72601_CR19","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1038\/sdata.2018.178","volume":"5","author":"TJ Pollard","year":"2018","unstructured":"Pollard, T. J. The eicu collaborative research database, a freely available multi-center database for critical care research. Sci. Data 5, 1\u201313 (2018).","journal-title":"Sci. Data"},{"key":"72601_CR20","doi-asserted-by":"publisher","DOI":"10.1038\/s41597-022-01899-x","volume":"10","author":"AE Johnson","year":"2023","unstructured":"Johnson, A. E. Mimic-iv, a freely accessible electronic health record dataset. Sci. Data 10, 1 (2023).","journal-title":"Sci. Data"},{"key":"72601_CR21","doi-asserted-by":"publisher","first-page":"e93","DOI":"10.1016\/S2589-7500(23)00226-1","volume":"6","author":"AA Soltan","year":"2024","unstructured":"Soltan, A. A. A scalable federated learning solution for secondary care using low-cost microcomputing: privacy-preserving development and evaluation of a covid-19 screening test in uk hospitals. Lancet Digit. Health 6, e93\u2013e104 (2024).","journal-title":"Lancet Digit. Health"},{"key":"72601_CR22","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-023-00805-y","volume":"6","author":"J Yang","year":"2023","unstructured":"Yang, J., Soltan, A. A., Eyre, D. W., Yang, Y. & Clifton, D. A. An adversarial training framework for mitigating algorithmic biases in clinical machine learning. NPJ Digit. Med. 6, 55 (2023).","journal-title":"NPJ Digit. Med."},{"key":"72601_CR23","doi-asserted-by":"publisher","first-page":"884","DOI":"10.1038\/s42256-023-00697-3","volume":"5","author":"J Yang","year":"2023","unstructured":"Yang, J., Soltan, A. A., Eyre, D. W. & Clifton, D. A. Algorithmic fairness and bias mitigation for clinical machine learning with deep reinforcement learning. Nat. Mach. Intell. 5, 884\u2013894 (2023).","journal-title":"Nat. Mach. Intell."},{"key":"72601_CR24","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-022-00614-9","volume":"5","author":"J Yang","year":"2022","unstructured":"Yang, J., Soltan, A. A. & Clifton, D. A. Machine learning generalizability across healthcare settings: insights from multi-site covid-19 screening. NPJ Digit. Med. 5, 69 (2022).","journal-title":"NPJ Digit. Med."},{"key":"72601_CR25","doi-asserted-by":"publisher","first-page":"e0235424","DOI":"10.1371\/journal.pone.0235424","volume":"15","author":"S Sheikhalishahi","year":"2020","unstructured":"Sheikhalishahi, S., Balaraman, V. & Osmani, V. Benchmarking machine learning models on multi-centre eicu critical care dataset. Plos ONE 15, e0235424 (2020).","journal-title":"Plos ONE"},{"key":"72601_CR26","unstructured":"Gupta, M. et al. An Extensive data processing pipeline for MIMIC-IV. In Proceedings of Machine Learning Research. (2022)."},{"key":"72601_CR27","doi-asserted-by":"crossref","unstructured":"Jang, J. et al. Knowledge unlearning for mitigating privacy risks in language models. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (2023).","DOI":"10.18653\/v1\/2023.acl-long.805"},{"key":"72601_CR28","unstructured":"Guo, C., Goldstein, T., Hannun, A. & Van Der Maaten, L. Certified data removal from machine learning models. In Proceedings of the International Conference on Machine Learning. (2020)."},{"key":"72601_CR29","doi-asserted-by":"publisher","unstructured":"Goel, S. et al. Towards adversarial evaluations for inexact machine unlearning. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2201.06640 (2022).","DOI":"10.48550\/arXiv.2201.06640"},{"key":"72601_CR30","doi-asserted-by":"publisher","first-page":"1957","DOI":"10.52202\/075280-0095","volume":"36","author":"M Kurmanji","year":"2023","unstructured":"Kurmanji, M., Triantafillou, P., Hayes, J. & Triantafillou, E. Towards unbounded machine unlearning. Adv. Neural Inf. Process. Syst. 36, 1957\u20131987 (2023).","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"72601_CR31","doi-asserted-by":"publisher","unstructured":"Shamsian, A., Shaar, E., Navon, A., Chechik, G. & Fetaya, E. Go beyond your means: Unlearning with per-sample gradient orthogonalization. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2503.02312 (2025).","DOI":"10.48550\/arXiv.2503.02312"},{"key":"72601_CR32","doi-asserted-by":"crossref","unstructured":"Shokri, R., Stronati, M., Song, C. & Shmatikov, V. Membership inference attacks against machine learning models. In Proceedings 2017 IEEE Symposium on Security and Privacy (SP). (2017).","DOI":"10.1109\/SP.2017.41"},{"key":"72601_CR33","doi-asserted-by":"crossref","unstructured":"Fredrikson, M., Jha, S. & Ristenpart, T. Model inversion attacks that exploit confidence information and basic countermeasures. In Proceedings of the 22nd ACM SIGSAC Conference on Computer and Communications Security. (2015).","DOI":"10.1145\/2810103.2813677"},{"key":"72601_CR34","doi-asserted-by":"publisher","unstructured":"Zhou, Z. et al. Model inversion attacks: A survey of approaches and countermeasures. https:\/\/doi.org\/10.48550\/arXiv.2411.10023 (2024).","DOI":"10.48550\/arXiv.2411.10023"},{"key":"72601_CR35","unstructured":"Van der Maaten, L. & Hinton, G. Visualizing data using t-sne. J. Mach. Learn. Res. 9, 2579\u20132605 (2008)."},{"key":"72601_CR36","unstructured":"Lundberg, S.M., and Lee, S.I. A unified approach to interpreting model predictions. In Proceedings Advances in Neural Information Processing Systems Syst. (2017)."},{"key":"72601_CR37","doi-asserted-by":"publisher","first-page":"1247","DOI":"10.1016\/j.clsr.2018.08.006","volume":"34","author":"E Politou","year":"2018","unstructured":"Politou, E., Michota, A., Alepis, E., Pocs, M. & Patsakis, C. Backups and the right to be forgotten in the gdpr: An uneasy relationship. Comput. Law Security Rev. 34, 1247\u20131257 (2018).","journal-title":"Comput. Law Security Rev."},{"key":"72601_CR38","doi-asserted-by":"crossref","unstructured":"Politou, E. et al. The \u201cright to be forgotten\u201d in the gdpr: implementation challenges and potential solutions. in Privacy and Data Protection Challenges in the Distributed Era (2022).","DOI":"10.1007\/978-3-030-85443-0_4"},{"key":"72601_CR39","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-023-41703-x","volume":"14","author":"J Zhou","year":"2023","unstructured":"Zhou, J. A unified method to revoke the private data of patients in intelligent healthcare with audit to forget. Nat. Commun. 14, 6255 (2023).","journal-title":"Nat. Commun."},{"key":"72601_CR40","doi-asserted-by":"crossref","unstructured":"Lin, S., Zhang, X., Susilo, W., Chen, X., and Liu, J. Gdr-gma: Machine unlearning via direction-rectified and magnitude-adjusted gradients. In Proceedings of the 32nd ACM International Conference on Multimedia. (2024).","DOI":"10.1145\/3664647.3680775"},{"key":"72601_CR41","doi-asserted-by":"crossref","unstructured":"Feng, Q. et al. Fg-oriu: Towards better forgetting via feature-gradient orthogonality for incremental unlearning. In Proceedings of the IEEE\/CVF International Conference on Computer Vision. (2025).","DOI":"10.1109\/ICCV51701.2025.00190"},{"key":"72601_CR42","doi-asserted-by":"crossref","unstructured":"Dehdashtian, S., Sadeghi, B. & Boddeti, V.N. Utility-fairness trade-offs and how to find them. In Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition. (2024).","DOI":"10.1109\/CVPR52733.2024.01144"},{"key":"72601_CR43","doi-asserted-by":"crossref","unstructured":"Cooper, A.F., Abrams, E. & Na, N. Emergent unfairness in algorithmic fairness-accuracy trade-off research. In Proceedings of the 2021 AAAI\/ACM Conference on AI, Ethics, and Society. (2021).","DOI":"10.1145\/3461702.3462519"},{"key":"72601_CR44","doi-asserted-by":"publisher","first-page":"137","DOI":"10.1504\/IJSN.2015.071829","volume":"10","author":"G Ateniese","year":"2015","unstructured":"Ateniese, G. Hacking smart machines with smarter ones: How to extract meaningful data from machine learning classifiers. Int. J. Secur. Netw. 10, 137\u2013150 (2015).","journal-title":"Int. J. Secur. Netw."},{"key":"72601_CR45","doi-asserted-by":"crossref","unstructured":"Ganju, K., Wang, Q., Yang, W., Gunter, C.A. & Borisov, N. Property inference attacks on fully connected neural networks using permutation invariant representations. In Proceedings of the 2018 ACM SIGSAC Conference on Computer and Communications Security. (2018).","DOI":"10.1145\/3243734.3243834"},{"key":"72601_CR46","unstructured":"Tram\u00e8r, F., Zhang, F., Juels, A., Reiter, M.K. & Ristenpart, T. Stealing machine learning models via prediction {APIs}. In Proceedings 25th USENIX security symposium (USENIX Security 16). (2016)."},{"key":"72601_CR47","doi-asserted-by":"crossref","unstructured":"Wang, B. & Gong, N.Z. Stealing hyperparameters in machine learning. In Proceedings 2018 IEEE symposium on security and privacy (SP). (2018).","DOI":"10.1109\/SP.2018.00038"},{"key":"72601_CR48","unstructured":"Kone\u010dn\u00fd, J. et al. Federated learning: Strategies for improving communication efficiency. In NIPS Workshop on Private Multi-Party Machine Learning (2016)."},{"key":"72601_CR49","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s41666-020-00082-4","volume":"5","author":"J Xu","year":"2021","unstructured":"Xu, J. Federated learning for healthcare informatics. J. Healthcare Inf. Res. 5, 1\u201319 (2021).","journal-title":"J. Healthcare Inf. Res."},{"key":"72601_CR50","unstructured":"Vaswani, A. et al. Attention is all you need. In Proceedings Advances in Neural Information Processing Systems 30 (2017)."},{"key":"72601_CR51","unstructured":"Hardt, M., Price, E. & Srebro, N. Equality of opportunity in supervised learning. In Proceedings Advances in Neural Information Processing Systems (2016)."}],"container-title":["Nature Communications"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41467-026-72601-7","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41467-026-72601-7.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41467-026-72601-7.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,8]],"date-time":"2026-07-08T23:03:07Z","timestamp":1783551787000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41467-026-72601-7"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,4]]},"references-count":51,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["72601"],"URL":"https:\/\/doi.org\/10.1038\/s41467-026-72601-7","relation":{},"ISSN":["2041-1723"],"issn-type":[{"value":"2041-1723","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,4]]},"assertion":[{"value":"1 July 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"17 April 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"4 May 2026","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":"6009"}}