ASH 2025 | Generative AI-powered clinical decision support in germline predisposition to myeloid neoplasms

Germline predisposition in myeloid neoplasm is a topic that is gaining a lot of attention because recently, in the last decade I will say, the availability of techniques that are able to screen our entire genome identified a lot of traits that are inherited and linked to specific myeloid neoplasms. We can say that in all comers, MDS and AML, 5 to 10% have inherited traits. So this means that it’s a non-negligible fraction and knowing so, it’s important because of the clinical repercussions that we have if we suspect that there’s like a familial trait segregating in the patient’s family. Because, for instance, if the patient needs a transplant, we must rule out that the donor, if it’s a relative, has the same mutation that perhaps is not being brought to light by an actual disease. Therefore, it’s important to have the tools to understand and how to identify these patients is critical. So together with a Spanish group of colleagues, Adrian Mosquera-Arguera and Andres Jerez, we created a chatbot that is a sort of like our series for hematologists that need tips on how to identify or, if identified, how to interpret molecular results. And this chatbot is very easy to use. You can ask questions in multiple languages or even ask what to do with this clinical vignette. For instance, a patient with a DDX41 mutation, 62-year-old. And of course, we vet all the literature that is fed into the system, to the chatbot, so that the answer that the clinician gets is approved by an expert on germline predisposition. So we think that this tool can be helpful, especially for the hematologists practicing in centers that do not have a high expertise on this and want to know something like off-the-shelf, ready-to-use knowledge regarding germline predisposition to myeloid malignancies.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.