Title Data Linkage of Traffic Crash Datasets and Hospital Datasets for Investigating Influential factors Affecting Injury Severity
Year 2017
Degree Master
School Transportation Science, Tamkang University
Author
Summary The National Police Agency of the Ministry of the Interior defines three injury levels of traffic accident victims: dead within 24 hours after the accident, injured, and not injured. The definition of death is that the victim passed away within 24 hours after the accident took place. The World Health Organization defines death in a traffic accident as a death within 30 days of the accident, and the definition of death in most of the developed countries is the same. Among all traffic accident victims in Taiwan, the injury level for 60% victims was grouped as the injured level by the definition of National Police Agency. It is necessary to distinguish between injuries for the large amount of injured victims.To define the detailed injury severity levels for crash victims, this study linked the National Police Agency traffic crash datasets from 2012 to 2015 to the Ministry of Health and Welfare datasets. The victims’ levels of injuries were grouped into five levels, including death within 30 days, hospitalized, emergency, injured, and not injured. The data analysis results show that, on average, the number of accident victims dead within 24 hours is approximately 1,900 persons (0.3%); the number of accident victims dead within 30 days is approximately 3,000 persons (0.5%); the number of accident victims hospitalized is approximately 48,500 persons (7.9%); and the number of accident victims emergency is approximately 20 million (36.7%)annually. In addition, due to the seriousness of motorcycle accidents in Taiwan, this study use Logistic regression model to explore the important factors that affect the death of motorcycle diver in different death definitions. This study also established an ordered logit model to explore the factors that affect the injury severity levels of motorcycle riders in traffic accidents. The results, based on the model, show that to reduce motorcycle riders’ level of injury, attention should be payed to the following traffic violations: drunk driving, speeding, and not wearing a helmet while riding a motorcycle. This study provides insights into the problems with the data and the data analysis process so that errors in the data analysis of the linked datasets can be avoided. The results of this study serve as a reference for safety-related authorities and follow-up studies.
Hashtags
View count: 243