EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Fast Interpretable Greedy-Tree Sums

Yan Shuo Tan, Chandan Singh, Keyan Nasseri, Abhineet Agarwal, James Duncan, Omer Ronen, Matthew Epland iD, Aaron Kornblith iD, Bin Yu

Proceedings of the National Academy of Sciences · 18.02.2025 · 122 (7)

DOI: 10.1073/pnas.2310151122 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 14.02.2025. Verfasst von Yan Shuo Tan und 8 weiteren Autor:innen. Beteiligte Einrichtung: Department of Statistics and Data Science, National University of Singapore, Singapore 119077, Republic of Singapore. Hauptthemen laut MEDLINE: Decision Trees, Machine Learning, Clinical Decision-Making. Die Arbeit wurde bisher 17-mal zitiert (Crossref) und nennt 79 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National University of Singapore (NUS), DOD | USN | ONR | U.S. Naval Research Laboratory (NRL), NSF | CISE | Division of Information and Intelligent Systems (IIS).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Modern machine learning has achieved impressive prediction performance, but often sacrifices interpretability, a critical consideration in high-stakes domains such as medicine. In such settings, practitioners often use highly interpretable decision tree models, but these suffer from inductive bias against additive structure. To overcome this bias, we propose Fast Interpretable Greedy-Tree Sums (FIGS), which generalizes the Classification and Regression Trees (CART) algorithm to simultaneously grow a flexible number of trees in summation. By combining logical rules with addition, FIGS adapts to additive structure while remaining highly interpretable. Experiments on real-world datasets show FIGS achieves state-of-the-art prediction performance. To demonstrate the usefulness of FIGS in high-stakes domains, we adapt FIGS to learn clinical decision instruments (CDIs), which are tools for guiding decision-making. Specifically, we introduce a variant of FIGS known as Group Probability-Weighted Tree Sums (G-FIGS) that accounts for heterogeneity in medical data. G-FIGS derives CDIs that reflect domain knowledge and enjoy improved specificity (by up to 20% over CART) without sacrificing sensitivity or interpretability. Theoretically, we prove that FIGS learns components of additive models, a property we refer to as disentanglement. Further, we show (under oracle conditions) that tree-sum models leverage disentanglement to generalize more efficiently than single tree models when fitted to additive regression functions. Finally, to avoid overfitting with an unconstrained number of splits, we develop Bagging-FIGS, an ensemble version of FIGS that borrows the variance reduction techniques of random forests. Bagging-FIGS performs competitively with random forests and XGBoost on real-world datasets.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Decision TreesMachine LearningClinical Decision-MakingAlgorithmsHumans

Stichwörter der Autor:innen

CARTXGBoostclinical decision instrumentinterpretabilityrandom forest

Autor:innen und Institutionen

  • Department of Statistics and Data Science, National University of Singapore, Singapore 119077, Republic of Singapore.
  • Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, CA 94720.
  • Microsoft Research, Redmond, Washington, WA 98052.
  • Statistics Department, University of California, Berkeley, CA 94720.
  • Graduate Group in Biostatistics, University of California, Berkeley, CA 94720.
  • Verana Health, San Francisco, CA 94107.

ORCID-Profile: Matthew Epland, Aaron Kornblith

Förderung

  • National University of Singapore (NUS)
  • DOD | USN | ONR | U.S. Naval Research Laboratory (NRL)
  • NSF | CISE | Division of Information and Intelligent Systems (IIS)
  • National Science Foundation (NSF)
  • HHS | NIH | Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)
  • NICHD NIH HHS
  • NSF | CISE | Division of Computing and Communication Foundations (CCF)
  • NSF | MPS | Division of Mathematical Sciences (DMS)

Zitieren

APA

Tan, Y. S., Singh, C., Nasseri, K., Agarwal, A., Duncan, J., Ronen, O., Epland, M., Kornblith, A., & Yu, B. (2025). Fast Interpretable Greedy-Tree Sums. Proceedings of the National Academy of Sciences, 122(7). https://doi.org/10.1073/pnas.2310151122

Vancouver

Tan YS, Singh C, Nasseri K, Agarwal A, Duncan J, Ronen O, et al. Fast Interpretable Greedy-Tree Sums. Proceedings of the National Academy of Sciences. 2025;122(7). doi:10.1073/pnas.2310151122

BibTeX
@article{Tan2025,
  author = {Tan, Yan Shuo and Singh, Chandan and Nasseri, Keyan and Agarwal, Abhineet and Duncan, James and Ronen, Omer and Epland, Matthew and Kornblith, Aaron and Yu, Bin},
  title = {Fast Interpretable Greedy-Tree Sums},
  journal = {Proceedings of the National Academy of Sciences},
  year = {2025},
  volume = {122},
  number = {7},
  doi = {10.1073/pnas.2310151122}
}
Kostenlos starten →