Related Experiment Video
Updated: May 10, 2025
03:14
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
462
Enhancing Malignancy Detection and Tumor Classification in Pathology Reports: A Comparative Evaluation of Large
Sabrina B Neururer1,2, Hasan Taha1,2, Helmut Muehlboeck1
1Department of Clinical Epidemiology, Tirol Kliniken GmbH, Innsbruck, Tirol, Austria.
Studies in Health Technology and Informatics
|April 24, 2025
Summary
Large language models (LLMs) significantly improve cancer data classification accuracy and efficiency. These advanced AI tools show great promise for enhancing public health monitoring and clinical decisions in oncology.
Area of Science:
- Computational pathology
- Artificial intelligence in oncology
- Health informatics
Background:
- Cancer registries need precise and efficient malignancy documentation.
- Current manual methods are labor-intensive and prone to errors.
Purpose of the Study:
- To assess the efficacy of large language models (LLMs) in classifying malignancies.
- To evaluate LLM performance in identifying tumor types from pathology reports.
Main Methods:
- A synthetic dataset of 227 pathology reports was utilized.
- Four LLMs and a score-based algorithm were benchmarked against expert-labeled data.
Main Results:
- LLMs, specifically GPT-4o and Llama3.3, achieved high sensitivity and specificity.
- LLMs substantially outperformed traditional algorithms in malignancy detection and tumor classification.
Conclusions:
- LLMs offer a significant advancement in the accuracy and efficiency of cancer data classification.
- LLMs show potential for improving public health surveillance and clinical decision-making in cancer care.
