EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Machine learning methods for estimating personalized treatment effects—insights on validity from two large trials

Hongruyu Chen, Helena Aebersold, Milo Alan Puhan iD, Miquel Serra-Burriel

American Journal of Epidemiology · 05.08.2026 · 195 (8) · 2263-2272

DOI: 10.1093/aje/kwag065 ↗

Kurz gefasst

Fachartikel in der Zeitschrift American Journal of Epidemiology, online erschienen am 20.03.2026. Verfasst von Hongruyu Chen und 3 weiteren Autor:innen. Beteiligte Einrichtung: Epidemiology, Biostatistics and Prevention Institute, University of Zurich, Zurich, Switzerland. Hauptthemen laut MEDLINE: Machine Learning, Precision Medicine, Stroke. Die Arbeit nennt 57 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: SNSF (Swiss National Science Foundation), SBFI (Swiss State Secretariat for Education and Innovation).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Machine learning (ML) methods have the potential to improve precision medicine by estimating personalized treatment effects. However, formal validation of these methods remains limited, leaving their reliability in empirical settings largely uncertain. In this study, we evaluated the internal and external validity of 17 causal heterogeneity ML methods—including metalearners, tree-based methods, and deep learning methods—using data from two large randomized controlled trials: the International Stroke Trial (n = 19 435) and the Chinese Acute Stroke Trial (n = 21 106). We assessed performance using three visual-based metrics and three quantitative metrics. Our analysis found that none of the ML methods consistently demonstrated reliable performance, neither internal nor external. Heterogeneous treatment effects estimated from training data failed to generalize to the test data, even in the absence of distribution shifts. These results raise concerns about the current applicability of ML models in precision medicine and highlight the need for more robust validation techniques to ensure generalizability.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Machine LearningPrecision MedicineStrokeHumansTreatment Effect HeterogeneityReproducibility of ResultsRandomized Controlled Trials as TopicPredictive Learning Models

Stichwörter der Autor:innen

causal machine learningindividualized treatment effectsvalidation

Autor:innen und Institutionen

  • Epidemiology, Biostatistics and Prevention Institute, University of Zurich, Zurich, Switzerland.
  • Department of Epidemiology, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland, United States.

ORCID-Profile: Milo Alan Puhan

Förderung

  • SNSF (Swiss National Science Foundation)
  • SBFI (Swiss State Secretariat for Education and Innovation)

Zitieren

APA

Chen, H., Aebersold, H., Puhan, M. A., & Serra-Burriel, M. (2026). Machine learning methods for estimating personalized treatment effects—insights on validity from two large trials. American Journal of Epidemiology, 195(8), 2263-2272. https://doi.org/10.1093/aje/kwag065

Vancouver

Chen H, Aebersold H, Puhan MA, Serra-Burriel M. Machine learning methods for estimating personalized treatment effects—insights on validity from two large trials. American Journal of Epidemiology. 2026;195(8):2263-2272. doi:10.1093/aje/kwag065

BibTeX
@article{Chen2026,
  author = {Chen, Hongruyu and Aebersold, Helena and Puhan, Milo Alan and Serra-Burriel, Miquel},
  title = {Machine learning methods for estimating personalized treatment effects—insights on validity from two large trials},
  journal = {American Journal of Epidemiology},
  year = {2026},
  volume = {195},
  number = {8},
  pages = {2263--2272},
  doi = {10.1093/aje/kwag065}
}
Kostenlos starten →