Knowledge graph–based thought: a knowledge graph–enhanced LLM framework for pan-cancer question answering
2025

Knowledge Graph-Based Thought for Cancer Question Answering

Sample size: 405 publication 10 minutes Evidence: high

Author Information

Author(s): Yichun Feng, Lu Zhou, Chao Ma, Yikai Zheng, Ruikun He, Yixue Li

Primary Institution: Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences

Hypothesis

Can integrating knowledge graphs with large language models improve the accuracy of biomedical question answering?

Conclusion

The KGT framework significantly enhances the accuracy and utility of large language models in the biomedical field.

Supporting Evidence

  • The KGT framework reduces factual errors in reasoning by utilizing verifiable information from knowledge graphs.
  • KGT can assist in predicting drug resistance by analyzing relevant biomarkers and genetic mechanisms.
  • The framework demonstrates strong adaptability and performs well across various open-source large language models.
  • KGT is the first knowledge graph question answering benchmark in the field of biomedicine.

Takeaway

This study shows that combining smart databases with AI can help doctors find better answers to cancer-related questions.

Methodology

The study developed a framework that integrates large language models with knowledge graphs to improve question answering in biomedicine.

Limitations

The dataset may not cover all potential use cases and the system currently does not perform fuzzy matching.

Digital Object Identifier (DOI)

10.1093/gigascience/giae082

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication