Speaker
Harrison Reeder, PhD
Long COVID is a chronic condition following acute SARS-CoV-2 infection and is characterized by a variety of persistent and potentially disabling symptoms. In this talk, we will take a deep dive into how various statistical challenges have been addressed in the analysis of adult and pediatric data from the Researching COVID to Enhance Recovery (RECOVER) Initiative. First, we will discuss the study's resource-efficient design in which certain assessments are only performed in a subset of individuals, with sampling based on values of auxiliary variables. This sampling occurs repeatedly over multiple visits until an individual is selected, yielding a sample drawn with complex time-varying selection probabilities dependent on auxiliary variable trajectories, and temporal dependence between the timing of sampling and timing of outcome measurement. Second, we will discuss cross-sectional and longitudinal clustering of complex data types, including social determinants of health and repeated measurements of self-reported symptom data. These approaches include novel Bernoulli mixture modeling and latent Markov models with sparse negative-unlabeled data for characterizing Long COVID progression.
Zoom webinar link: https://duke.zoom.us/j/93145784765?pwd=Mo1ZldD1uvZWOUDjdeuwXv2ZO6riyd.1
This event is being cross-promoted by the NC BERD Consortium, a collaboration of the CTSA-funded BERD cores at UNC-Chapel Hill, Wake Forest University School of Medicine, and Duke University School of Medicine.