Journal Pre-proof A machine learning model of response to hypomethylating agents in myelodysplastic syndromes in

: 35 Hypomethylating agents (HMA) prolong survival and improve cytopenias in individuals 36 with higher-risk myelodysplastic syndrome (MDS). Only 30-40% of patients, however, respond 37 to HMAs, and responses may not occur for more than 6 months after HMA initiation. We 38 developed a model to more rapidly assess HMA response by analyzing early changes in 39 patients’ blood counts. Three institutions’ data were used to develop a model that assessed 40 patients’ response to therapy 90 days after the initiation using serial blood counts. The model 41 was developed with a training cohort of 424 patients from 2 institutions and validated on an 42 indepdent cohort of 90 patients. The final model achieved an area under the receiver operating 43 characteristic curve (AUROC) of 0.79 in the train/test group and 0.84 in the validation group. 44 The model provides cohort-wide and individual-level explanations for model predicitons, and 45 model certainty can be interrogated to gauge the reliability of a given prediction.