Online Program

Predicting Katrina effect on low birth weight in gulf coast based on hybrid model of PLS regression and GM (1, N)

Tuesday, November 5, 2013

Chau-Kuang Chen, Ed.D., School of Graduate Studies, Meharry Medical College, Nashville, TN
Patricia Matthews-Juarez, PhD, School of Medicine, Meharry Medical College, Nashville, TN
Aiping Yang, PhD, Industrial Engineering, Beijing Union University, Beijing, China
Colette Davis, M.S., School of Graduate Studies and Research, Meharry Medical College, Hermitage, TN
This paper applies the hybrid model of partial least squares (PLS) regression and GM (1, N) to study the relationship between adverse health outcome and its risk factors. PLS is a dimension reduction tool that has the ability to handle collinearity issue. The grey model is part of the state-of-the-art Grey Theory System originally developed by Julong Deng in 1989 to deal with uncertain and incomplete information known as grey. It is a curve fitting model that is prominent for its use in predicting future values with fewer observations. Using these sophisticated techniques, the effect of Hurricane Katrina on low birth weight babies for African American and White women was examined in Louisiana, Mississippi and Alabama. The risk factors used in this study included the number of total primary care physicians, unemployment rate, poverty rate for all ages, median household income, total birth rate, percent of African American mothers aged 15-19, and percent of White mothers aged 15-19, state group (Gulf Coast states vs. other U.S. states), and timeline (2003-2004 before Katrina vs. 2006 after Katrina. The GM model performed accurate predictions for low birth weight trend for African American and White women. Data presented between 2007 and 2010 showed low birth weight for African American women exhibited a sharp increasing trend in Gulf Coast while the trend for White women demonstrated a slight increase. The empirical findings could provide valuable information to policy makers in their efforts to make effective policies for improving health disparity.

Learning Areas:

Biostatistics, economics
Public health or related research

Learning Objectives:
Identify risk factors affecting low birth weigh in Gulf Coast area Demonstarte prediction accuracy of combined paritial least squares regression and grey model

Keyword(s): Health Disparities, Infant Mortality

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I am a Biostatistics professor specializing in time series and artificial intelligence modeling techniques
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.