An artificial intelligence (AI) platform that analyzed bone marrow biopsies and digital disease images was able to differentiate cases of prefibrotic primary myelofibrosis (pre-PMF) and essential thrombocythemia (ET) with 92.3% accuracy.
Prefibrotic myelofibrosis and essential thrombocythemia, which are both types of myelofproliferative neoplasms, are historically difficult to differentiate.
However, as AI continues to make strides in cancer care, it is essential that patients and clinicians alike are understanding that while the goal is to make these programs representative of the general population — in this case, patients with myelofibrosis or essential thrombocythemia — it might be biased one way or another, explained Dr. Andrew Srisuwananukorn, of The Ohio State University Comprehensive Cancer Center.
Srisuwananukorn presented his findings at the 2023 American Society of Hematology Annual Meeting. At the conference, he sat down with CURE® to discuss what patients need to know about this AI model.
Transcript
I think there are two aspects that we really should be considering as we develop these AI algorithms. Number one, it’s important for us to make sure that (regarding) the training, it’s important for us to understand that the algorithm was developed on a patient cohort. And we really want that patient cohort to be representative of the general population, it might accidentally learn a feature of the cohort that we didn’t need to that has no basis in biology. So, it’s important that our algorithm is representative of all patient cohorts of at risk populations.
For more news on cancer updates, research and education, don’t forget to subscribe to CURE®’s newsletters here.
