Immune checkpoint inhibitors (ICIs) can be lifesaving for cancer patients, but they can also cause immune-related side effects (irAEs) that can affect nearly every organ in the body. The frequency and severity of these side effects are not well understood, making it hard for doctors to manage them effectively. Current methods to track these side effects are manual and inefficient.
Researchers from Mass General Brigham have used a large language model (LLM) to better identify these side effects in hospitals. The AI tool found common and severe side effects like colitis, hepatitis, pneumonitis, and myocarditis caused by ICIs. It was more accurate than the traditional International Classification of Disease (ICD) codes and even found additional cases that were missed by manual checks. The results were published in The Journal of Clinical Oncology.
“The LLM showed higher accuracy in detecting irAEs and identified more cases that were missed by manual review,” said Dr. Kerry Reynolds, Director of the Severe Immunotherapy Complications Program at Mass General Cancer Center. “As a free and open-source model, this tool can help other institutions create similar databases and foster collaboration in new ways.”
The study analyzed 10 years of data from patients who were hospitalized after receiving ICI therapy. The AI model consistently achieved over 90% accuracy. Dr. Reynolds noted that the tool is easy to use, requires minimal computing power, and can be run on a local computer, making it accessible for smaller hospitals to contribute to research in this field.
Source: massgeneralbrigham
