Please join us for a Joint GWU/CTSI-CN/Washington DC VA Informatics Seminar taking place Wednesday, June 25th, 11:00am - 12:00pm Eastern Daylight Time. Dr. Joseph L. Goulet of Yale University will present:
Can we Apply More Statistical Methods in Informatics?
The objective of this presentation is to discuss whether some statistical methods are or can be used in informatics, and if so, what research may benefit from their application. This stems from work on comparing results of statistical parameter estimates to impact factors from machine learning models. There are multiple assumptions in statistics that should be assessed before and after applying analytic methods, such as goodness-of-fit and residuals. Other basic but essential considerations include independence of observations, distribution of variables, particularly the outcome variable, inclusion of relevant covariates, functional form, and model choice. In this presentation I will present some of these methods and ask attendees to discuss their potential utility and application within the context of machine learning.
Dr. Goulet is Professor Emeritus of Emergency Medicine at the Yale University School of Medicine, and Consultant for Observational Studies, Yale Center for Analytical Sciences (YCAS). He is interested in health services research focusing on Veterans with psychiatric and medical comorbidity. His specific interests include research on: the prognostic significance of comorbidity among people with HIV, the ‘clustering’ of comorbid diseases and conditions, and the treatment of chronic pain among patients with current or pre-existing substance use disorders. Dr. Goulet received a PhD in Epidemiology from Yale University, and an MS in statistics from Southern Connecticut State University.