Dr. Robert Aseltine and Dr. Kun Chen will offer a short course entitled “Practical Integrative Statistical Learning: Recent Developments and Case Studies” at the 2017 New England Statistics Symposium (NESS), to be held at UConn, April 21-22.
This short course focuses on practical predictive modeling and statistical learning techniques for analyzing large-scale heterogeneous data. In many fields, measurements of several distinct yet interrelated sets of characteristics pertaining to a single set of subjects and possibly collected from an array of sources, has become increasingly common. For example, individual health data may come from insurance claims, pharmacy visits, clinical records, patient surveys, and government statistics. The availability of such complex data makes tackling many fundamental scientific problems possible through “integrative statistical learning”, which is undergoing exciting development and is pushing for a genuine refinement of the conventional multivariate learning toolkit. In this course, several classes of interpretable predictive modeling techniques for simultaneous dimension reduction, feature construction and model estimation will be introduced. Practical case studies in health informatics regarding suicide prevention, drug abuse, race and ethnic disparities in health outcomes, etc, together with examples from insurance, finance and industrial engineering will be discussed. The course consists of 4 modules: 1) overview of integrative statistical learning and health informatics; 2) dimension reduction techniques with case studies; 3) integrative predictive modeling techniques with case studies; 4) More recent developments on data fusion and demonstrations with R.