Student seminar series
Lily Zou
PhD Candidate, University of Waterloo
Room: M3 3127
Using Real-World Data to Study Complex Disease Processes
Cohort studies and disease registries can provide valuable information on patient disease progression and response to treatment. However, the complex interplay between acute phases of disease activity, disease progression, clinic visits, and treatment changes create challenges in extracting real-world evidence about treatment effects. Disease activity is often associated with disease progression which in turn influences treatment decisions, thus introducing confounding by indication. Moreover, information on disease activity and progression is gathered upon encounters with the healthcare system, the times of which can be related to disease activity. In this talk, we formulate joint multistate models for the disease, activity markers, treatment and observation processes which can address these complexities. We also quantify the biases that can arise from fitting simpler models. We demonstrate how such models be leveraged to estimate marginal treatment effects using G-computation.