G. Gómez Melis, A. Toloba López-Egea, K. Langohr
Interval censoring has gained increasing attention due to advancements in automated data collection for longitudinal studies. Traditionally, it has been widely applied in survival analysis, where the time to an event is only known to occur within a random time interval. A novel application arises in metabolomics, where metabolite concentrations measured via liquid chromatography cannot always be precisely quantified due to limits of detection (LoD) and quantitation (LoQ).
In this study, we investigate the association between composition of multiple metabolites and anthropometric and biochemical parameters using a Generalized Linear Model (GLM). We consider the case where the predictor of interest is the sum of several metabolite concentrations, which is interval-censored within a range defined by the LoD and LoQ of its components. Our focus is on proposing and evaluating both standard GLM and quantile residuals to assess GLM assumptions, comparing their performance in this context.
Keywords: Survival analysis; Interval Censoring; GLM
Scheduled
Biostatistics I
June 13, 2025 11:00 AM
MR 3