An Approach to Evaluating Treatment Selection-Biomarker for Observational Studies

• Activity Number: 25 - Medical Devices and Diagnostics Speed Session
• Type: Contributed
• Date/Time: Sunday, August 8, 2021 : 1:30 PM to 3:20 PM
• Sponsor: Section on Medical Devices and Diagnostics
• Author(s): Hulya Kocyigit* and Kevin K Dobbin
• Companies: University of Georgia and University of Georgia
• Keywords: treatment selection marker; observational studies; causal inference; trimming ; weighting; lung cancer

Abstract

Identifying biomarkers that can guide patients to specific treatments they can benefit from is a compelling goal.There is increased interest in creating biomarkers to assess treatment response and improve patient treatment decisions in randomized trials. Also, treatment selection biomarker can reduce medical cost, develop clinical outcome, and lessen the burden of side effects from potential toxicities.However, there is a limited statistical framework for evaluating biomarkers from observational studies in the literature.This paper aims to recommend a suitable method by designing and testing a treatment selection process based on data obtained from non-randomized trial settings.Our study was motivated by a lung cancer dataset from an observational study among patients that evaluated the effect of adjuvant chemotherapy in preventing progression to severe cancer disease.Besides, we will examine whether the causal inference is necessary for identifying treatment selection biomarkers, and we will discuss the types of causal inference techniques that should be preferred.