Files
Abstract
ABSTRACTBackground
The Centers for Disease Control and Prevention (CDC) recommends sharing science and data faster. This study focuses on forecasting motor vehicle (MV) related deaths in the United States (US) to share forecasted data faster while examining how demographic factors—such as age, sex, race/ethnicity, and geographic location—affect these fatalities.
Methods
This dissertation consists of two studies. The first study utilizes new and improved forecasting procedures to identify highly accurate techniques that provide estimates, bridging the gap created by the lack of timely data. The second study employs two regression models to describe the associations between demographic factors and MV fatalities. The project utilizes the National Vital Statistics System (NVSS) motor vehicle mortality data from July 2020 to June 2024 (N = 175,600 deaths) in the US.
Results
The study achieves a high average forecast accuracy of 98% and provides predictions for 12 months (July 2024-June 2025) based on demographic factors. Among subgroups, the analysis reveals significantly higher reported MV deaths among males, by race white individuals, by urbanization in large central metro areas, by age within the age group of 25-44 years, and during August to October months.
Conclusions
The high forecast accuracy achieved through the Exponential Smoothing Model (ESM) demonstrates its potential to enhance the timeliness of the provisional data (by eliminating a one-year data lag) and help inform effective traffic safety strategies. The findings highlight the urgent need for targeted interventions to reduce MV fatalities and safeguard public health and safety. Targeted interventions should focus on young adult males aged 25–44, mainly white individuals in large central metro areas, with enhanced enforcement, education, and prevention campaigns intensified during the high-risk months of August to October, also improved road signage, better lighting, and changes in road layout.
INDEX WORDS: Motor vehicle accident; forecast; predict; trend; time series; demographic factors; negative binomial regression.