Informed Treatment Decisions; Model-Based Meta-Analysis + Machine Learning = Accurately Ranking Treatment Effectiveness + Predicting New Trials

Session Description

This session will address the application of comparative effectiveness in informing patients and physician decisions on the best treatment options in the absence of data from randomized head-to-head (H2H) superiority trials. Through a real-case of a new treatment for psoriatic arthritis, we will show novel ways of combining model-based meta-analysis (MBMA) and machine learning methods to rank treatment options in a robust, cost-effective and timely way and present such information in a format that can be easily digested by the patients and physicians. We will demonstrate how these methodologies can be combined to confidently predict the probability of success of future H2H trials that could be run to provide additional evidence to further guide patient/physician treatment choices.

In addition to the case example of psoriatic arthritis, we will provide brief vignette examples of MBMA supporting discovery, late nonclinical, early-phase clinical studies, pivotal studies and outcomes research in a Ted-talk style. Novel examples will be illustrated such as using MBMA for target selection, highlighting opportunity of one target over another; MBMA estimation of the clinical therapeutic window from nonclinical pharmacology studies; MBMA optimal design of Phase 1 or 2 studies ensuring the correct patients, assessments, and assessment timing is performed; and MBMA patient stratification to ensure the landscape of therapeutic options is applied to the right patient at the right time. The objectives of these vignettes are to illustrate nonstandard use of MBMA throughout the development of a therapy and to enable the patient to understand therapeutic options across a variety of measures simultaneously.

Learning Objectives

1. Understand why we need Model-Based Meta-Analysis (MBMA) to accurately and fairly rank multiple treatments from early discovery to patient-focused materials.
2. Demonstrate how machine learning methodologies were used to assess the predictive performance of candidate MBMA models to ensure the Probability of Success (PoS) of potential head-to-head trials were sound.
3. See, though a real case study, how patient/physician treatment decisions can be guided by advanced modeling and simulation methods that accurately rank all treatments regimens and used to design H2H trials with a high PoS.

Session Schedule

Introduction
William Denney, PhD and Ghada Ahmed, PhD

A Novel Comparative Effectiveness Analysis to Inform Ranking Available Treatments in Psoriatic Arthritis and Confidently Determine Probability of Success for Future Superiority Trials
Alan Maloney, MSc, PhD and Ghada Ahmed, PhD

Atypical Use of MBMA to Support All Stages of Development and Patient Care
William Denney, PhD

Fireside Chat and Audience Q&A