Skip to main content
[Postponed] Optimization of Behavioral and Biobehavioral Interventions
Add to Calendar 2020-04-14T20:00:00 2020-04-14T21:00:00 UTC [Postponed] Optimization of Behavioral and Biobehavioral Interventions 116 Henderson Building, University Park, PA
Start DateTue, Apr 14, 2020
4:00 PM
End DateTue, Apr 14, 2020
5:00 PM
Presented By
Linda Marie Collins, Penn State Clinical and Translational Science Institute (CTSI), Distinguished Professor, Human Development and Family Studies

A Bayesian Approach to Making Decisions Based on the Results of a Factorial Optimization Trial

Event Series:

Behavioral and biobehavioral interventions play a critically important role in many areas of health. A few examples (out of many possible) include interventions aimed at weight loss, prevention and treatment of substance misuse, prevention of HIV transmission, and improvement of compliance with the diabetes treatment regimen. To have maximal public health impact, interventions must be not only effective, but also affordable, readily implementable, and scalable—i.e., capable of having wide reach. The multiphase optimization strategy (MOST) is an innovative, engineering-inspired framework for developing, optimizing, and evaluating behavioral and biobehavioral interventions with the objective of attaining high public health impact. In MOST, an optimization phase of research precedes evaluation by randomized control trial (RCT). In this optimization phase, a randomized, powered optimization trial estimates the individual and combined effects of intervention components. Then, based on the results of the optimization trial, investigators decide which components to include in the optimized intervention. The objective of this decision-making is to identify the set of intervention components that yields the best expected outcome while remaining affordable, implementable, and scalable. The current methods of decision-making in the optimization phase of MOST are based on classical hypothesis testing, a frequentist approach. However, Bayesian methods are better equipped to answer the questions that motivate decision-making, such as “What is the probability that a particular set of intervention components yields the best outcome (e.g. the biggest reduction in substance misuse)?”

In this presentation we will provide a brief introduction to intervention optimization. We will then present some very new work on a Bayesian approach to decision-making when decisions are based on the results of a factorial optimization trial. We hypothesize that this approach will more successfully identify the set of components that constitutes the optimal intervention.