72: Addressing Measurement Errors and Zero-Inflation in Functional Joint Linear Quantile Regression

Music City Center 

Wearable devices collect time-varying biobehavioral data, offering opportunities to investigate how behaviors influence health outcomes. However, these data often contain measurement error and excess zeros (due to nonwear, sedentary behavior, or connectivity issues), each characterized by subject-specific distributions. Current statistical methods fail to address these issues simultaneously and estimate the quantiles separately. We propose a novel estimation process that accounts for subject-specific measurement error and subject-specific time-varying zero-inflation probabilities in a functional joint linear quantile regression model. Our approach integrates a linear mixed model to adjust for measurement error and a maximum-likelihood estimation for zero inflation. Through extensive simulations, we demonstrate that our approach significantly improves estimation accuracy over methods that only address measurement error. When applied to a childhood obesity study, our approach achieves superior predictive performance and provides bootstrap-based inference, offering insights into how various predictors influence body mass index at different quantile levels.

Functional Data

Joint Quantile Regression

Measurement Error

Wearable Devices-Collected Data

Zero-Inflation 

Statistics Without Borders