Finding causal relationships is the most profound motivation behind research in various disciplines. However, our toolbox of computational methods for causal inference has remained limited for a long time. Better computational methods for causal inference can change the way that we conduct research and have the potential to revolutionize science by expediting scientific discoveries and cutting costs. Share your thoughts and experiences with us here and on Yale Causal Inference and Reasoning Working Group (YCIR).
At Yale PCCSM Analysis hub, we have developed a method called 'Causal Inference Using Composition of Transactions' (CICT). CICT uses machine learning on large-scale observational datasets to infer causal relationships. Moreover, we created a pipeline of conventional reasoning models and methods to evaluate the results of CICT. CICT can be used to efficiently and accurately infer underlying system in a large and complex transition or relevance networks.
Details on the method and results can be found here: https://lnkd.in/ecnMFUu
We have had a very productive time since October 2017 producing novel etiological insights using CICT causal inference pipeline. We had four abstracts accepted to different conferences, six papers-in-progress, one finished and more in the pipeline.
In collaboration with four teams we will present our latest findings at ATS Conference 2018 - American Thoracic Society: