A0336
Title: Bayesian joint modeling for longitudinal patient-reported outcomes and survival data
Authors: Dae Jin Lee - IE University (Spain) [presenting]
Cristina Galan Arcicollar - BCAM - Basque Center for Applied Mathematics (Spain)
Josu Najera Zuloaga - University of the Basque Country (Spain)
Danilo Alvares - University of Cambridge (United Kingdom)
Abstract: In recent years, there has been a growing interest in understanding how longitudinal biomarkers influence the occurrence of significant events, such as mortality. In this context, two outcomes from the same subject are simultaneously observed: Repeated measures and time-to-event data. The inherent association between these outcomes has led to the development of a joint modeling framework, which simultaneously analyzes these two outcomes to capture their interdependence. At the same time, modern healthcare research is increasingly focusing on patient-centered approaches. Patient-reported outcomes (PROs) serve as valuable tools for informing clinicians about patients' perspectives on their health, symptoms, and quality of life. However, the statistical properties of PROs are not always adequately addressed during analysis, leading to biased estimates and suboptimal predictions. A Bayesian joint modeling approach is proposed for longitudinal PRO data and survival outcomes to address these challenges. This approach incorporates the distributional features of PROs to better fit the data. In addition, dynamic predictions of time-to-event outcomes are provided according to the model. It allows for personalized prognostic updates, a powerful tool for clinicians to enhance decision-making and improve patient care.
