ELM (Ensemble of Language Models) dramatically accelerates tumor group classification from pathology reports, reducing a 900-hour manual process to automated analysis.

• Combines multiple fine-tuned small language models with large language models in a novel ensemble approach
• Effectively extracts critical data from unstructured pathology reports
• Significantly reduces human workload for population-based cancer registries
• Demonstrates how specialized AI ensembles can solve complex healthcare classification challenges

This research represents a breakthrough for medical data processing, enabling more efficient cancer surveillance and research by automating the extraction of vital diagnostic information from clinical documents.

ELM: Ensemble of Language Models for Predicting Tumor Group from Pathology Reports