Oct 7
Math/Stats Colloquium: Abby Loe, University of Michigan
Title: A Wrinkle in Time (to Event): Inverse Probability Weighted Pseudo-observations for Analysis of Alternating Recurrent Events
Abstract: Alternating recurrent events, where subjects experience two potentially correlated event types over time, are common in clinical, healthcare, social and behavioral studies. Often there is a primary event of interest that, when triggered, initiates a period of treatment and recovery measured via a secondary time-to-event. For example, cancer patients can experience repeated blood clotting emergencies that require hospitalization, people with alcohol use disorder can have periods of addiction and sobriety, or medical interns experience periods of high or low mood. Potential censoring of the data requires special handling, and overlaying this are the missing at-risk periods when individuals are being treated for the condition that initiated the primary event. In this talk, we develop a framework for regression analysis of censored alternating recurrent events that uses a random forest inverse probability weighting strategy to avoid bias in the analysis of the primary time-to-event due to informative missingness from those in the alternate secondary state. The proposed regression model estimates τ-restricted mean time to primary outcome of interest while taking into account complexities of censored/missing data. Simulations show good performance of our method when the alternate times-to-event are either independent or correlated. We provide an application arising from a mobile health study, evaluating the impact of a just-in-time adaptive intervention on the mental state of carers of traumatic brain injury patients.
Note from the organizers: Abby Loe graduated from Carleton in 2021. She is currently a PhD candidate in biostatistics at University of Michigan.