CancerLLM: A Large Language Model in Cancer Domain
This provides a more efficient and robust tool for clinical research and practice in cancer, though it is incremental as it adapts existing methods to a specific domain.
The researchers tackled the lack of a specialized large language model for cancer phenotyping and diagnosis by developing CancerLLM, a 7-billion-parameter model pre-trained on clinical notes and pathology reports, which achieved state-of-the-art F1 scores of 91.78% on phenotyping extraction and 86.81% on diagnosis generation, with an average improvement of 9.23% over existing LLMs.
Medical Large Language Models (LLMs) have demonstrated impressive performance on a wide variety of medical NLP tasks; however, there still lacks a LLM specifically designed for phenotyping identification and diagnosis in cancer domain. Moreover, these LLMs typically have several billions of parameters, making them computationally expensive for healthcare systems. Thus, in this study, we propose CancerLLM, a model with 7 billion parameters and a Mistral-style architecture, pre-trained on nearly 2.7M clinical notes and over 515K pathology reports covering 17 cancer types, followed by fine-tuning on two cancer-relevant tasks, including cancer phenotypes extraction and cancer diagnosis generation. Our evaluation demonstrated that the CancerLLM achieves state-of-the-art results with F1 score of 91.78% on phenotyping extraction and 86.81% on disganois generation. It outperformed existing LLMs, with an average F1 score improvement of 9.23%. Additionally, the CancerLLM demonstrated its efficiency on time and GPU usage, and robustness comparing with other LLMs. We demonstrated that CancerLLM can potentially provide an effective and robust solution to advance clinical research and practice in cancer domain