Using Causal Inference to Estimate What-if Outcomes for Targeting Treatments

Individuals often have heterogeneous outcomes after interventions. As a result, clinicians constantly ask themselves, given a patient’s history, what would happen to the patient’s clinical trajectory if they were given one treatment versus another. In order to target care, estimating how outcomes or responses to treatments will vary across individuals is critical. However, in practice it is often unknown how the patient’s signals will change in response to a treatment until that treatment is actually administered. Furthermore, even if we have observed data on the patient for one treatment, the counterfactual, i.e., what would have happened to the patient if the doctor had made a different choice, is unobserved.