Research Maps for Integrating Research and Planning Experiments
Causality is a central concept for both basic science and clinical medicine. In the last few decades, we have seen significant development of mathematical formalisms for modeling causality. Despite the existence of robust and expressive formalisms for causal modeling, such formalisms are surprisingly underused by biologists seeking to identify causal mechanisms and by clinicians seeking to understand the etiology of disease. MII is thus working to adapt state-of-the-art causal-discovery methods so that experts in these domains can leverage causal models as integral parts of their research and clinical workflows. The field of causal discovery seeks to learn causal models from data and/or domain knowledge. Moving beyond such pure causal discovery, our research group is also working to define a new paradigm of experiment planning that is based on the mathematics of causal models. By grounding the process of experiment planning in the formalisms of causal models, we are developing metrics that can be used to compare the causal information gain of individual experiments, thus allowing scientists to prioritize experiments by their ability to minimize the uncertainty in existing causal models. As part of this work, researchers in MII are collaborating with neuroscientist Alcino J. Silva on a project named ResearchMaps. Visit researchmaps.org to learn more.