A1044
Title: Flexible hazards and cure models for dynamic prediction based on longitudinal biomarker measurements
Authors: Xuelin Huang - University of Texas MD Anderson Cancer Center (United States) [presenting]
Can Xie - University of Texas MD Anderson Cancer Center (United States)
Ruosha Li - The University of Texas Health Science Center at Houston (United States)
Nicholas Short - University of Texas MD Anderson Cancer Center (United States)
Christopher Flowers - The University of Texas MD Anderson Cancer Center (United States)
Abstract: To optimize personalized treatment strategies, treat patients efficiently, and extend their survival times, it is critical to accurately predict patients' prognoses at all stages, from disease diagnosis to follow-up visits. The longitudinal biomarker measurements during these visits are essential for this prediction purpose. Patients' ultimate concerns are cure and survival. However, in many situations, there is no clear biomarker indicator for cure. A comprehensive joint model of longitudinal and survival data is proposed, incorporating proportions of potentially cured patients. Formulas are provided for predicting an individual's probabilities of future cure and survival at any time point based on their current biomarker history. The survival distributions in the model are specified through flexible hazard functions with the proportional hazards as a special case, allowing other patterns such as crossing hazard and survival functions. Simulations show that, with these comprehensive and flexible properties, the proposed model outperforms commonly used models in terms of predictive performance, measured by the time-dependent area under the curve (AUC) of the receiver operating characteristic and the Brier score. The use and advantages of the proposed model are illustrated by its application to a study of patients with chronic myeloid leukemia.
