Charles DiMaggio, PhD, Assistant Professor of Epidemiology, Columbia University, discusses the benefits of incorporating Bayesian hierarchical modeling into spatial analysis of public health data. He asserts that 21st century advances in statistics and computing allow the full appreciation and use of methods first described 2 centuries ago by Rev. Bayes. He also provides examples of how this form of spatial analysis can benefit disaster preparedness and response efforts. This course also covers the basic hierarchy of the Bayesian approach and conditional auto-regression. Participants in this course should have a basic understanding of geospatial analysis.
The purpose of this course is to:
- Provide an in depth understanding of how recent advances in statistics and computing allow for new forms of spatial analyses that can benefit first responders, epidemiologists and social science researchers.
- Bayesian hierarchical modeling is a powerful tool for helping researchers answer difficult geospatial questions
- First responders, epidemiologists and social science researchers need to gain more familiarity with Bayesian hierarchical modeling, to better understand how it can be incorprorated into planning, response, and research. There are free computer tools, such as WinBUGS, which can help facilitate an understanding of the Bayesian approach.
- Bayesian hierarchical modeling is a challenging concept, but the learning curve is not as high as most people assume. Professor DiMaggio describes how we can incorporate Bayesian thinking into our everyday lives.
Public Health Emergency Preparedness Capabilities:
- Public Health Surveillance and Epidemiologic Investigation
- Emergency Public Information and Warning
- Information Sharing
Preparedness and Emergency Response Learning Center (CDC):
- 2.3 Report information potentially relevant to the identification and control of an emergency through the chain of command.
- 3.2 Contribute expertise to the development of emergency plans.
- 3.3 Participate in improving the organization's capacities (including, but not limited to program, plans, policies, laws, and workforce training).
At the end of this course, you will be able to:
- Describe how Bayesian statistics and computing allow for new forms of spatial analyses that can benefit first responders, epidemiologists and social science researchers
- Identify and explain the relevance of conditional auto-regression
- Define and summarize the similarities and differences between frequentist and Bayesian statistical approaches
- Summarize the purpose and importance of unstructured heterogeneity in spatial data
- Describe the utility of Ripley's K function
- Course Material - Spatial Approaches to Disaster Epidemiology
- Satisfaction Survey
- Spatial Approaches to Disaster Epidemiology: Dr. Snow Meets the Rev. Bayes