The MIT researchers have spent the past few years developing the Synthetic Survival Control causal inference framework, which enables them to answer complex “when-if” questions when using available data is statistically challenging. Their approach estimates when a target event happens if a certain intervention is used.

In this paper, the researchers investigated an aggressive cancer called nodal mature T-cell lymphoma, and whether a certain prognostic marker led to worse outcomes. The marker, TTR12, signifies that a patient relapsed within 12 months of initial therapy.

They applied their framework to estimate when a patient will die if they have TTR12, and how their survival trajectory would be different if they do not have this prognostic marker.