In 2017, North Carolina experienced one of the worst influenza seasons on record, with 12,396 citizens testing positive for the disease and 389 flu-related deaths. We predict that socioeconomic factors have a considerable effect on flu contraction and death rates in North Carolina. Understanding which areas are affected most by these societal factors would help with targeted and more effective drug delivery to prevent similarly devastating flu seasons.

With population data for the 100 North Carolina counties from the 2012 to 2016 American Community Survey 5 year estimates, certain socioeconomic attributes were selected: population, education level, age, gender, poverty level, race, flu caused deaths and health insurance coverage. Linear and logistic regression models were then yielded with the support vector neural network, Weka 3-8. Using the regression models, an analysis was conducted to fit and normalize flu death to find rate. A cross validation of 10 folds ultimately output statistical analysis measures within a linear regression, logistic regression and F-measure. Based on analysis of the data, the selected socioeconomic attributes can help predict influenza deaths. Poverty and unemployment rates of counties were the greatest factors, and persons under the age of 5 and those over 65 showed higher flu death rates. A heat map analysis utilizing ArcGIS was also conducted to visualize these socioeconomic factors with respect to distance from local hospitals. This map would be beneficial for targeted vaccinations and prevention of a future flu outbreak.