Online Program Home
My Program

Abstract Details

Activity Number: 355 - Contributed Poster Presentations: Biopharmaceutical Section
Type: Contributed
Date/Time: Tuesday, July 30, 2019 : 10:30 AM to 12:20 PM
Sponsor: Biopharmaceutical Section
Abstract #308018
Title: The Statistics of Synthetically-Controlled Clinical Trials
Author(s): Aaron Smith and Charles K. Fisher*
Companies: Unlearn.AI and Unlearn.AI
Keywords:
Abstract:

The previous decade’s symbiotic advancement in computational infrastructure and machine learning have caused an efflorescence of powerful capabilities for modeling clinical data. In particular, it has become feasible to build unsupervised generative models which model the progression of particular diseases under standard-of-care treatment. Such models are capable of generating synthetic patient records which could be used to supplement or replace control arms of traditional randomized controlled trials (RCT). We take for granted that such a model exists, and consider the statistical implications of replacing the control arm of an RCT with a synthetic control. We first demonstrate show how statistical tests and associated type I and II errors change when a simulation is used in place of the control arm. Secondly we demonstrate how these errors can be controlled given knowledge of the degree to which the synthetic control distribution accords with the true control distribution.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program