Constructing Stabilized Dynamic Surveillance Rules for Optimal Monitoring Schedule

• Activity Number: 3 - New Developments and Challenges for Dynamic Individualized Treatments
• Type: Invited
• Date/Time: Sunday, August 7, 2022 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Xinyuan Dong and Yingye Zheng and Yingqi Zhao*
• Companies: Amazon Inc and Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center
• Keywords: Dynamic surveillance rules; active surveillance; personalized medicine; benefit-risk tradeoff

Abstract

Dynamic surveillance rules (DSRs) are sequential surveillance decision rules informing the monitoring schedules in clinical practice, which can adapt overtime according to a patient's evolving characteristics. In clinical applications, it is desirable to identify optimal stabilized DSRs, where the parameters indexing the DSRs are shared across different times. We propose a new criterion for DSRs accouting for benefit-cost tradeoff during disease surveillance. We develop two methods to estimate the stabilized DSRs optimizing the proposed criterion, and establish asymptotic properties for the estimated parameters. The first approach estimates the optimal decision rules at every stage via regression modeling, and then estimates the stabilized DSRs via a classification procedure. The second approach proceeds by optimizing a relaxation of the empirical objective. Simulations and a data nalaysis using the Canary Prostate Active Surveillance Study (PASS) are conducted to demonstrate the superior performances of the proposed methods.