Click here to download PDF copy of presentation

Speaker: Gary Summers (GSK)

Abstract:

We’ve all heard that statistic’s mantra, “Correlation does not equal causation.” However, causal inference, a new set of statistical techniques, can identify causal relationships from data. Amazingly, it can derive experimental results and counterfactuals from observational data. DAGs, directed acyclic graphs, similar to influence diagrams, are the heart of causal inference. Using ten years of publicly available clinical trial data, Intelligencia and GSK are creating a DAG and studying causation in clinical trial designs, a prelude measuring the impact of decision choices, such as how enriching trials with patient selection biomarkers impact a trial’s probability of success. This presentation will introduce causal inference, DAGs, and our early results.