EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Minimizing and quantifying uncertainty in AI-informed decisions: Applications in medicine

Samuel D Curtis iD, Sambit Panda iD, Adam Li iD, Haoyin Xu, Yuxin Bai iD, Itsuki Ogihara, Eliza O'Reilly, Yuxuan Wang, Lisa Dobbyn, Maria Popoli, Janine Ptak, Nadine Nehme und 22 weitere

Proceedings of the National Academy of Sciences · 26.08.2025 · 122 (34)

DOI: 10.1073/pnas.2424203122 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 20.08.2025. Verfasst von Samuel D Curtis und 33 weiteren Autor:innen. Beteiligte Einrichtung: Department of Pharmacology and Molecular Sciences, Johns Hopkins University School of Medicine, Baltimore, MD 21205. Hauptthemen laut MEDLINE: Artificial Intelligence. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 88 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: HHS | NIH (NIH), Benjamin Baker Endowment, Oncology Core CA.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

AI is now a cornerstone of modern dataset analysis. In many real world applications, practitioners are concerned with controlling specific kinds of errors, rather than minimizing the overall number of errors. For example, biomedical screening assays may primarily be concerned with mitigating the number of false positives rather than false negatives. Quantifying uncertainty in AI-based predictions, and in particular those controlling specific kinds of errors, remains theoretically and practically challenging. We develop a strategy called multidimensional informed generalized hypothesis testing (MIGHT) which we prove accurately quantifies uncertainty and confidence given sufficient data, and concomitantly controls for particular error types. Our key insight was that it is possible to integrate canonical cross-validation and parametric calibration procedures within a nonparametric ensemble method. Simulations demonstrate that while typical AI based-approaches cannot be trusted to obtain the truth, MIGHT can be. We apply MIGHT to answer an open question in liquid biopsies using circulating cell-free DNA (ccfDNA) in individuals with or without cancer: Which biomarkers, or combinations thereof, can we trust? Performance estimates produced by MIGHT on ccfDNA data have coefficients of variation that are often orders of magnitude lower than other state of the art algorithms such as support vector machines, random forests, and Transformers, while often also achieving higher sensitivity. We find that combinations of variable sets often decrease rather than increase sensitivity over the optimal single variable set because some variable sets add more noise than signal. This work demonstrates the importance of quantifying uncertainty and confidence—with theoretical guarantees—for the interpretation of real-world data.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Artificial IntelligenceHumansUncertaintyNeoplasmsLiquid BiopsyAlgorithms

Stichwörter der Autor:innen

biomarkersbiomedical assayscancer screeninghypothesis testingpredictive modeling

Autor:innen und Institutionen

  • Department of Pharmacology and Molecular Sciences, Johns Hopkins University School of Medicine, Baltimore, MD 21205.
  • Department of Oncology, the Sidney Kimmel Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD 21205.
  • The Ludwig Center, Johns Hopkins University School of Medicine, Baltimore, MD 21205.
  • The Sol Goldman Pancreatic Cancer Research Center, Johns Hopkins University School of Medicine, Baltimore, MD 21205.
  • Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD 21287.
  • Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD 21218.

ORCID-Profile: Samuel D Curtis, Sambit Panda, Adam Li, Yuxin Bai, Lan T Ho-Pham, Joshua T Vogelstein

Förderung

  • HHS | NIH (NIH)
  • Benjamin Baker Endowment
  • Oncology Core CA
  • NCI NIH HHS
  • Federal Government | DHAC | National Health and Medical Research Council (NHMRC)

Zitieren

APA

Curtis, S. D., Panda, S., Li, A., Xu, H., Bai, Y., Ogihara, I., O'Reilly, E., Wang, Y., Dobbyn, L., Popoli, M., Ptak, J., Nehme, N., Silliman, N., Tie, J., Gibbs, P., Ho-Pham, L. T., Tran, B. N. H., Tran, T. S., Nguyen, T. V., …, & Douville, C. (2025). Minimizing and quantifying uncertainty in AI-informed decisions: Applications in medicine. Proceedings of the National Academy of Sciences, 122(34). https://doi.org/10.1073/pnas.2424203122

Vancouver

Curtis SD, Panda S, Li A, Xu H, Bai Y, Ogihara I, et al. Minimizing and quantifying uncertainty in AI-informed decisions: Applications in medicine. Proceedings of the National Academy of Sciences. 2025;122(34). doi:10.1073/pnas.2424203122

BibTeX
@article{Curtis2025,
  author = {Curtis, Samuel D and Panda, Sambit and Li, Adam and Xu, Haoyin and Bai, Yuxin and Ogihara, Itsuki and O'Reilly, Eliza and Wang, Yuxuan and Dobbyn, Lisa and Popoli, Maria and Ptak, Janine and Nehme, Nadine and Silliman, Natalie and Tie, Jeanne and Gibbs, Peter and Ho-Pham, Lan T and Tran, Bich N H and Tran, Thach S and Nguyen, Tuan V and Irajizad, Ehsan and Goggins, Michael and Wolfgang, Christopher L and Wang, Tian-Li and Shih, Ie-Ming and Fader, Amanda and Lennon, Anne Marie and Hruban, Ralph H and Bettegowda, Chetan and Gilbert, Lucy and Kinzler, Kenneth W and Papadopoulos, Nickolas and Vogelstein, Bert and Vogelstein, Joshua T and Douville, Christopher},
  title = {Minimizing and quantifying uncertainty in AI-informed decisions: Applications in medicine},
  journal = {Proceedings of the National Academy of Sciences},
  year = {2025},
  volume = {122},
  number = {34},
  doi = {10.1073/pnas.2424203122}
}
Kostenlos starten →