Metraipolis › archive › dc_a57d045a2f2c

Memorisation bias in medical AI — arXiv:2609.17223

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.17223
License (as archived)
CC0-1.0
Archived
2026-10-15T19:30:58.660Z
SHA-256 of the archived text
4ecab4c10fb5701dafdac5c4c1bc12f0b4b8ff73ff8e60667f979e35de4141e0
Stable URL of this page
https://metraipolis.com/documento/dc_a57d045a2f2c

How to cite

Memorisation bias in medical AI — arXiv:2609.17223 — arXiv (en), https://arxiv.org/abs/2609.17223 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a57d045a2f2c (Metraipolis archive dc_a57d045a2f2c, sha256 4ecab4c10fb5701dafdac5c4c1bc12f0b4b8ff73ff8e60667f979e35de4141e0)
BibTeX
@misc{dca57d045a2f2c,
  title = {Memorisation bias in medical AI — arXiv:2609.17223},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.17223},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a57d045a2f2c; sha256 4ecab4c10fb5701dafdac5c4c1bc12f0b4b8ff73ff8e60667f979e35de4141e0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.17223. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Medical AI models hold immense potential to improve patient outcomes, but they are also known to unintentionally memorise individual records from their training datasets. While such memorisation has been linked to targeted privacy attacks, its consequences for clinical deployment, where patients may be assessed by a model that saw their historical data during training, remain poorly understood. Here we show that predictions on a patient's unseen future data can change significantly if a model observed that same patient's anonymised historical data during training, a phenomenon we term "memorisation bias". We demonstrate that this bias exists across diverse data modalities and model architectures, and over prolonged time spans: in some cases, memorisation bias persists on future records acquired decades after the historical records used for training. Moreover, in simulated prospective deployment, memorisation bias has asymmetric effects on the diagnostic accuracy of returning data contributors. When a patient returned with a de novo condition absent from their historical records in the training dataset, diagnostic sensitivity decreased significantly compared to an otherwise identical model not trained on their historical data. Conversely, when their health state was unchanged, both sensitivity and specificity were significantly inflated. Our findings reveal a previously uncharacterised risk in medical AI that arises when a model is deployed on patients who contributed to its training data. This exposes a shortcoming of current model development practice: the de-identification measures designed to protect patients' privacy make it difficult to identify returning contributors and exclude them from the AI-assisted interpretation of their own future data. Mitigating memorisation risks may thus require changes to current model training and deployment protocols.

Tema degli esploratori: governance digitale e beni comuni (retro-catalogo arXiv: annunci del 2026-09-16 in cs.CY).
Categorie arXiv: Machine Learning (cs.LG) ; Computers and Society (cs.CY).
Pagina del lavoro: https://arxiv.org/abs/2609.17223 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T18:39:08.628Z dalla pagina https://arxiv.org/catchup/cs.CY/2026-09-16?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).