Online Program

Return to main conference page
Saturday, February 16
Sat, Feb 16, 8:00 AM - 9:15 AM
St. James Ballroom
Poster Session 3 and Continental Breakfast

Predictive Models of Health Care Expenditure: Penalized Regression Approaches Among All Level Income Economies (303810)

View Presentation View Presentation

Ahmad Talafha, Western Michigan University 
*Emmanuel Thompson, Southeast Missouri State University 

Keywords: Penalization, LASSO, Elastic net, Adaptive Elastic net, Adaptive LASSO, Monte Carlo method, Health expenditure per capita

Over the last decades, healthcare systems globally have seen intensive developments and improvement. Also, there is a remarkable variation in healthcare spending across income economies. In the literature, gross domestic product (GDP) per capita has been recognized as a vital predictor of health expenditure. However, there is no consensus on which other variables may be linked to the outstanding largely unexplained variation in health expenditure. Therefore, the present study aims to identify key predictors of healthcare expenditure for low-income, lower-middle-income, upper-middle-income, and high-income economies using a variety of penalization methods. The Monte Carlo method was used to establish statistical significance of the penalized regression coefficients. The study was based on 2014 World Bank data with several development and economic indicators. Initial results identified the adaptive elastic net algorithm as critical in identifying important predictors as well as accurately estimating healthcare expenditure per capita (HEC) among low-income, upper-middle-income and high-income economies. The adaptive LASSO was the best for estimating HEC among lower-middle economies.