New PREBA framework improves surgical duration prediction using LLMs

Researchers have developed PREBA, a novel framework designed to improve the accuracy of surgical duration predictions. This system utilizes retrieval-augmented large language models (LLMs) combined with principal component analysis (PCA) weighting and Bayesian averaging. Unlike traditional supervised methods that require extensive labeled data, PREBA operates in a zero-shot manner, grounding LLM reasoning in institution-specific clinical evidence without task-specific fine-tuning. Tested on four real-world datasets with LLMs like Qwen3, DeepSeek-R1, and HuatuoGPT-o1, PREBA significantly outperforms standard zero-shot inference and achieves accuracy competitive with, and sometimes superior to, supervised machine learning approaches. AI

IMPACT This framework could enhance resource management in hospitals by providing more accurate surgical duration predictions, potentially reducing costs and improving patient flow.

RANK_REASON The cluster describes a research paper detailing a new framework and its evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Bayesian Averaging is Well-Temperated
• DeepSeek-R1
• HuatuoGPT-o1
• Kanxue Li
• principal component analysis
• Qwen3

• paper
• model release

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