Abstract: Thanks to the work of (Athey and Imbens 2016), recursive partitioning for heterogeneous causal effects, there is a new available tool to carry on in a very simple and thorough way a systematic exploration of heterogeneous causal effects. The method partitions the feature space into subspaces and the output is a tree of covariate-based treatment effects.  The potential applications of such a methodology are endless, especially for randomized control trials. Using ISTG data, we apply this novel methodology to the IST 1 study, where aspirin intake, subcutaneous heparin intake, both or neither were randomized after an acute ischemic stroke. We discover novel treatment-covariate interactions that were ignored when clinical practice guidelines for stroke were developed.

Bio: Researcher at the Centre for Research in Health and Economics (CRES). His main areas of interest are the empirical prediction models applied to medical and health, the development of models for the economic evaluation of new drugs and health technologies (for the realization of other evaluation studies and cost-effectiveness studies), and theoretical modeling in health and the drug economy. Degree in economics from the Pompeu Fabra University (UPF) and Master in Health Economics and Public Policy from the Barcelona Graduate School of Economics (GSE). He joined CRES to work on the project "The Economic impact of a hypothetical change rx-to-otc in Spain" in 2014. He worked at the consulting firm specializing in the pharmaceutical sector, performing the tasks of modeling and preparing of dossiers of value for the negotiation of prices and access to the market of new drugs in Latin America. Currently his collaboration in CRES continues in several research projects ranging from the development of theoretical models for the definition of the portfolio of Health benefits with public funding, until the development of empirical models for Health Technologies Assessment.