EHA 2025 | SmartCytoFlow: AI-driven flow cytometry for lymphocyte quantification and CAR T-cell monitoring

We’ve just developed a new artificial intelligence platform whose name is SmartCytoFlow. And in this technological platform we host new machine learning models for flow cytometry automation. So we were very interested in lymphocyte quantification and particularly CAR T-cell detection because we are seeing in the clinical setting that there are lots of requests for these determinations in patients undergoing CAR T-cell therapy and the numbers are rising very, very fast, so the number of determinations are rising very fast too and the inter-observer heterogeneity is going to be higher too. So what we did here is to develop a specific unsupervised machine learning model that can analyze raw flow cytometry data, automatically compensate for all the data, and remove debris, tablets, etc. And then we use some expert experience to assign those clusters to a specific lymphocyte population. And we were able to develop a tool that can automatically classify these lymphocytes along with the determination and quantification of anti-CD19 and anti-BCMA CAR T-cells. So by now this tool is fully automatic, it runs automatically in our hospital and we are obviously deserving of establishing new collaborations to test this tool in other hospitals.

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