Machine Learning from Clinical and Genomic Data with Marshfield Clinic
Marshfield Clinic was one of the first clinics in the world to have an electronic health record and biobank of DNA from consenting patients. From a 2004 discussion with Dr. Michael Caldwell, then Marshfield Clinic Research Foundation director, we began using state-of-the-art machine learning algorithms to analyze de-identified data at Marshfield to construct clinical predictive models. Examples include models to predict heart attack or various adverse drug reactions. This work led to NIH research grants collaborative with Marshfield on privacy and on drug events. It also contributed to an NLM-funded training grant, ICTR grant, and Wisconsin Genomics Initiative started by Dr. Humberto Vidaillet that also included MCW and UW-Milwaukee.