Development of a dynamic counterfactual risk stratification strategy for newly diagnosed acute myeloid leukemia patients treated with venetoclax and azacitidine

The objective of this study was to develop a flexible risk stratification strategy for Acute Myeloid Leukemia (AML) that is specific for venetoclax plus azacitidine (ven/aza), addresses real-world data (RWD) issues, and is also adaptable to different use cases. A series of tunable risk models (RMs) were generated from a dynamic counterfactual machine learning (ML) strategy. These utilized a range of features from diagnostic AML samples and were tested using objective metrics on a single institution cohort of 316 newly diagnosed patients treated with ven/aza.RM performance was tested using various model assumptions, data elements, and endpoints, and with applications to an external AML real world cohort (RWC). Favorable, Intermediate, and Adverse risk groups were identified in a series of ML-based RMs using different assumptions, for genetic only or genetic-plus-phenotypic data elements and with overall survival and event free survival as endpoints. Most RMs demonstrated equitable patient distribution (∼20%-40% in each risk group), significant separation between risk strata (Log-rank based P-values <0.001), and predictability computed by time-dependent survival AUC values of 0.60-0.70. Similar performance was observed when the proposed RM strategy was adapted and compared to the ELN22 using the external RWC. The proposed ML strategy addresses a variety of RWD considerations and is readily tunable through coding and parameter updates for different contexts and use case needs. This strategy represents a novel approach to developing more effective RMs for AML and possibly other diseases. Can an effective machine learning based risk stratification strategy be developed for Acute Myeloid Leukemia (AML) that address real world challenges and varying use cases? A RM strategy for AML patients treated with venetoclax/azacitidine was developed based on a dynamic counterfactual machine learning (ML) model. This strategy efficiently generated specific RMs that demonstrated robust performance with varying assumptions, data elements, and endpoints, and was adaptable to an external real world cohort dataset. These findings demonstrate that ML-based RMs are feasible and may have potential advantages over existing AML RM strategies in adapting to context dependent settings.