How ChatGPT can augment breast cancer care

Background and purpose: 2023 has witnessed an explosion in artificial intelligence technology with the public release of large-language models such as ChatGPT (OpenAI). This study explores the potential of ChatGPT, to augment breast cancer care by aiding summation, information gathering, and information dissemination. The aim is to evaluate the feasibility and effectiveness of ChatGPT in enhancing patient engagement and healthcare delivery in the context of breast cancer. Methods: Two clinicians utilised ChatGPT to aid in various tasks, including summarising radiology and pathology reports, generating GP letters, and personalising patient information. De-identified patient information was input into the ChatGPT service at https://chat.openai.com and outputs were analysed. Results: ChatGPT is able to quickly and effectively perform a range of language-based tasks. Summarisation of data, such as pathology and radiology reports, is particularly effective, especially when a proforma for output is given. Humanising shorthand notes, for the purpose of inter-clinician and patient communication, is also effective, however there are occasional issues with factual inaccuracies and wildly irrelevant responses. ChatGPT is able to make medical recommendations based on patient data, which are often largely correct, however this is reliant on accurate baseline training material and accurate interpretation of data. ChatGPT was also liable to "hallucinating" – the generation of fictional or speculative content – requiring a clinician to fact-check generated responses. Conclusions: ChatGPT can vastly speed up language-based tasks such as summarising and changing the tone of established patient data. This can be very useful in breast cancer care, however the inability to fact-check generated information means that human input is still required. References: 1. Sorin, V., Barash, Y., Konen, E. & Klang, E. Deep-learning natural language processing for oncological applications. Lancet Oncol. 21, 1553–1556 (2020). 2. Sorin, V., Barash Y., Konen E., Klang E. Large language models for oncological applications. J. Cancer Res. Clin. Oncol. https://doi.org/10.1007/s00432-023-04824-w.