Evaluating the Use of Large Language Model to Identify Top Research Priorities in Hematology

The field of hematology is highly progressive and dynamic requiring researchers to commit significant amounts of time and effort towards staying abreast of the most crucial research areas. As such, in this work we assess the potential of ChatGPT for identifying research priorities within five key topics in hematology: acute lymphocytic leukemia, immunotherapy, targeted therapy, hematopoietic stem cell transplantation, and acute myeloid leukemia. After ChatGPT was employed to generate specific research questions in these areas, a panel of seven experienced hematologists independently reviewed and rated resultant research questions based on four parameters: relevance, originality, clarity, and specificity on a scale of 1 to 5, with 5 denoting the highest score. Excitingly, the mean and median grades of the four parameters were all above 4, indicating that the hematologists strongly agreed that the problems generated by ChatGPT were generally highly specific, clear, relevant, and original. As such, although further work will clearly be required, we suggest our current study indicates that Large Language Models (LLMs), such as ChatGPT, may very well represent valuable new tools for more efficiently identifying and prioritizing impactful research questions in the field of hematology.