A recent study reveals that large language models (LLMs) can generate treatment recommendations for early-stage hepatocellular carcinoma (HCC) that align with clinical guidelines. However, these models struggle with more complex, advanced cases. The research led by Ji Won Han from The Catholic University of Korea and published in the open-access journal PLOS Medicine highlights the potential and limitations of AI in guiding cancer treatment.
Choosing the right treatment for liver cancer is inherently challenging. Physicians must consider various factors, including the stage of cancer, liver function, and patient comorbidities, alongside international treatment guidelines. To evaluate the effectiveness of LLMs in this context, the researchers compared treatment recommendations generated by three LLMs—ChatGPT, Gemini, and Claude—with actual treatment data from over 13,000 newly diagnosed HCC patients in South Korea.
The findings indicate that for patients with early-stage HCC, there is a significant correlation between the recommendations provided by LLMs and the treatments administered. Notably, higher alignment between AI suggestions and actual clinical practices was associated with improved survival rates. Conversely, in patients with advanced-stage disease, the study observed that increased agreement between LLM recommendations and actual treatments corresponded to worse survival outcomes.
Researchers noted that LLMs tend to prioritize tumor characteristics, such as size and number of tumors, while physicians focus more on liver function. This discrepancy suggests that while LLMs may assist in straightforward treatment decisions for early-stage cancer, they are not yet suitable for managing more complex cases requiring nuanced clinical judgment.
The authors of the study emphasized the cautious use of LLM advice, stating, “Our study shows that large language models can help support treatment decisions for early-stage liver cancer, but their performance is more limited in advanced disease. This highlights the importance of using LLMs as a complement to, rather than a replacement for, clinical expertise.”
As AI technology continues to evolve, its role in healthcare remains a topic of interest. The findings from this research underscore the importance of integrating AI tools with traditional clinical approaches to enhance patient care while ensuring that human expertise remains at the forefront of complex medical decisions.
This study represents a significant step in evaluating the clinical utility of AI in oncology. As healthcare professionals explore these technologies, the implications for patient outcomes, especially in the realm of liver cancer treatment, will be closely monitored.
