Machine learning models evaluated for imbalanced clinical data

A new study published on arXiv explores the effectiveness of various machine learning models for predicting critical care outcomes using imbalanced clinical data. Researchers evaluated six model families, including tree-based methods and foundation models, on the MIMIC-IV-ED and eICU databases. The findings indicate that while XGBoost performed best on the eICU dataset, TabPFN v2.6 and TabICL showed strong results on MIMIC-IV-ED, suggesting no single model universally dominates. Foundation models offer a promising efficiency-performance trade-off for resource-constrained clinical settings. AI

IMPACT Suggests foundation models offer a distinct efficiency-performance trade-off for clinical prediction tasks with imbalanced data.

RANK_REASON Academic paper detailing empirical study of machine learning models on clinical data. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Decision Tree
• MIMIC-IV-ED
• Random Forest
• TabICL
• TabNet
• TabPFN v2.6
• XGBoost

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →