Publication:
A framework with a cost estimation model and the identification of high vehicle crash locations with injured persons in Puerto Rico
A framework with a cost estimation model and the identification of high vehicle crash locations with injured persons in Puerto Rico
Authors
Bianchi-Santiago, Josie
Embargoed Until
Advisor
Valdés-Díaz, Didier M.
College
College of Engineering
Department
Department of Civil Engineering
Degree Level
Ph.D.
Publisher
Date
2020-05-27
Abstract
Según la Organización Mundial de la Salud (2018), 1.35 millones de personas mueren cada año y de 20 a 50 millones de personas resultan heridas o discapacitadas cada año en choques automovilísticos. La Oficina de Estadísticas de Transporte del Departamento de Transporte de EE. UU. informó 84 heridos por cada 100 millones de millas vehiculares, con más de 2.7 millones de personas heridas por accidentes automovilísticos y 1.9 millones por choques automovilísticos solo con heridos en 2018. Puerto Rico experimentó un promedio de 150,000 accidentes automovilísticos por año, con un total de más de 5,700 muertes y 441,300 heridos entre 2002 y 2017. En 2018, Puerto Rico tuvo aproximadamente 15,000 accidentes automovilísticos solo con heridos, con más de 21,800 personas heridas y una tasa de 24.1 heridos graves por 100 millones de millas por vehículo.
El costo asociado con la lesión por choque en Puerto Rico se basa en la versión del Manual de Seguridad de Carreteras 2010 y los datos de la Administración Federal de Carreteras y está representado por cinco niveles de escala de gravedad de heridos por choque llamado KABCO. El modelo se basa en un estudio de 2005 de la Administración Federal de Carreteras que utilizó el valor de dólares en 2001 que no incluía a Puerto Rico. Las agencias de transporte y policía de Puerto Rico solo usan tres tipos de heridos en su distinción de tipo de accidente: fatal, heridos y daños a la propiedad.
Para abordar adecuadamente los esfuerzos de seguridad vial en Puerto Rico, se desarrolló un modelo y análisis de estimación de costos de accidentes. Este proceso se basó en los gastos médicos y los costos asociados para cada tipo de choque revisando la escala de heridos de KABCO para choques automovilísticos solo con heridos en Puerto Rico. Y, para generar una zona alta de costo de choque versus una comparación de zona alta de frecuencia de choque para identificar ubicaciones de choque de alto riesgo en función del impacto del costo médico para la sociedad.
Como primer paso, se realizó un proceso de integración de registros con dos bases de datos: 1) La base de datos de accidentes mantenida por el Departamento de Transportación de Puerto Rico, que consta de todos los registros de accidentes policiales presentados por la Policía de Puerto Rico; y 2) la base de datos de reclamos de seguro de lesiones personales relacionadas con el tráfico que mantiene la Administración de Compensación de Accidentes de Automóviles. Se realizaron dos tipos de integración de registros: determinista y probabilístico.
Después del proceso de integración de registros, como un segundo paso, se realizó un análisis de conglomerados de k-medias con los datos del servicio médico de lesiones relacionadas con el tráfico para determinar si la categorización KABCO de tres niveles para lesiones relacionadas con el choque se ajusta a los datos de Puerto Rico. Como resultado, la mejor configuración de clúster o grupo que maximizó la distancia entre los grupos y minimizó la distancia dentro de los grupos fue tres grupos de severidad del choque.
El tercer paso fue el desarrollo de un modelo de estimación del costo médico del accidente para explorar la correlación entre las circunstancias del accidente y los costos médicos de la persona lesionada relacionada con el choque. Se implementó la regresión lineal con la transformación logarítmica natural de la variable dependiente para desarrollar el modelo de estimación del costo médico del choque.
Como paso final, se implementó una comparación de metodología de identificación de zonas altas de choque con el costo médico de las personas lesionadas por choques automovilísticos en Puerto Rico. Con los datos disponibles (datos del choque e información de la carretera), se implementaron y compararon cuatro medidas diferentes de rendimiento de la zonas altas de choques: 1) frecuencia promedio del choque; 2) frecuencia de choque promedio equivalente de solo daños a la propiedad con factores de ponderación; 3) Función de rendimiento de seguridad Jaramillo para la identificación de ubicaciones de altos choques; y 4) probabilidad de tipos de choque específicos que exceden la proporción umbral usando el método estadístico G*.
Esta investigación desarrolló un marco para los modelos de estimación del costo del choque solo con heridos, la estimación de los niveles de severidad solo por heridos de KABCO y la identificación de zonas de altos choques de choques de vehículos con personas lesionadas en Puerto Rico. La información presentada en este estudio proporciona una mejor comprensión de los costos médicos relacionados con los accidentes en Puerto Rico. Proporcionará información para que los gerentes de proyectos de transportación tomen decisiones basadas en evidencia relacionadas con las mejoras de seguridad vial.
According to the World Health Organization, 1.35 million persons die each year and between 20 to 50 million persons are injured or disabled each year in car crashes (2018). The USA Department of Transportation Bureau of Transportation Statistics reported 84 injuries per 100 million vehicle-miles, with more than 2.7 million vehicle-crash injured persons and 1.9 million for injury-only motor-vehicle crashes for 2018. Puerto Rico experienced an average of 150,000 vehicle crashes per year, with a total of more than 5,700 deaths and 441,300 injuries from 2002 to 2017. In 2018, Puerto Rico had approximately 15,000 injury-only motor-vehicle crashes, with more than 21,800 injured persons and a rate of 24.1 serious injuries per 100 million vehicle-miles. The cost associated with crash injury in Puerto Rico is based on the Highway Safety Manual 2010 version and the Federal Highway Administration data and is represented by five levels of crash-injury severity scale named KABCO. The model is based on a 2005 study of the Federal Highway Administration that used a 2001-dollar value that did not include Puerto Rico. Puerto Rico transportation and police agencies only use three types of injuries in their crash type distinction: fatal, injury, and property-damage-only. To adequately address road safety efforts in Puerto Rico, a crash-cost estimation model and analysis was developed. This process was based on medical expenses and associated costs for each type of crash by revising the KABCO injury scale for injury-only motor vehicle crashes on Puerto Rico. And, to generate a crash-cost hotspot versus crash-frequency hotspot comparison to identify high-risk crash locations based on the impact of medical cost to society. As a first step, a record linkage process was performed with two databases: 1) the crash database maintained by the PRDOT, consisting of all police crash records submitted by the Puerto Rico Police; and 2) the traffic-related personal injury insurance claim database maintained by the Automobile Accidents Compensation Administration. Two types of record linkage were performed: deterministic and probabilistic. After the record linkage process, as a second step, a k-means cluster analysis was performed with the medical service data from traffic-related injuries to ascertain if the three-level KABCO categorization for traffic-related injuries fits the Puerto Rico data. As a result, the best cluster or group configuration that maximized the distance among groups and minimized the distance within groups was three crash severity groups. The third step was the development of a medical crash-cost estimation model to explore the correlation between crash circumstances and traffic-related injured person medical costs. Linear regression with the natural logarithmic transformation of the dependent variable was implemented to develop the medical crash-cost estimation model. As a final step, a crash-hotspot identification methodology comparison was implemented with the medical cost of injured persons for car crashes in Puerto Rico. With the data available (crash data and roadway information), four different high crash location performance measures were implemented and compared: 1) average crash frequency; 2) equivalent property-damage-only average crash frequency with weighting factors; 3) Jaramillo safety performance function for the identification of high-crash locations; and 4) probability of specific crash types exceeding threshold proportion using the statistical method G*. This research developed a framework for injury-only crash-cost estimation models, KABCO injury-only severity levels estimation, and identification of high crash locations of vehicle crashes with injured persons in Puerto Rico. The information presented in this study provide a better understanding of the medical costs related to crashes in Puerto Rico. It could provide information for highway safety managers to make evidence-based decisions related to highway safety improvements.
According to the World Health Organization, 1.35 million persons die each year and between 20 to 50 million persons are injured or disabled each year in car crashes (2018). The USA Department of Transportation Bureau of Transportation Statistics reported 84 injuries per 100 million vehicle-miles, with more than 2.7 million vehicle-crash injured persons and 1.9 million for injury-only motor-vehicle crashes for 2018. Puerto Rico experienced an average of 150,000 vehicle crashes per year, with a total of more than 5,700 deaths and 441,300 injuries from 2002 to 2017. In 2018, Puerto Rico had approximately 15,000 injury-only motor-vehicle crashes, with more than 21,800 injured persons and a rate of 24.1 serious injuries per 100 million vehicle-miles. The cost associated with crash injury in Puerto Rico is based on the Highway Safety Manual 2010 version and the Federal Highway Administration data and is represented by five levels of crash-injury severity scale named KABCO. The model is based on a 2005 study of the Federal Highway Administration that used a 2001-dollar value that did not include Puerto Rico. Puerto Rico transportation and police agencies only use three types of injuries in their crash type distinction: fatal, injury, and property-damage-only. To adequately address road safety efforts in Puerto Rico, a crash-cost estimation model and analysis was developed. This process was based on medical expenses and associated costs for each type of crash by revising the KABCO injury scale for injury-only motor vehicle crashes on Puerto Rico. And, to generate a crash-cost hotspot versus crash-frequency hotspot comparison to identify high-risk crash locations based on the impact of medical cost to society. As a first step, a record linkage process was performed with two databases: 1) the crash database maintained by the PRDOT, consisting of all police crash records submitted by the Puerto Rico Police; and 2) the traffic-related personal injury insurance claim database maintained by the Automobile Accidents Compensation Administration. Two types of record linkage were performed: deterministic and probabilistic. After the record linkage process, as a second step, a k-means cluster analysis was performed with the medical service data from traffic-related injuries to ascertain if the three-level KABCO categorization for traffic-related injuries fits the Puerto Rico data. As a result, the best cluster or group configuration that maximized the distance among groups and minimized the distance within groups was three crash severity groups. The third step was the development of a medical crash-cost estimation model to explore the correlation between crash circumstances and traffic-related injured person medical costs. Linear regression with the natural logarithmic transformation of the dependent variable was implemented to develop the medical crash-cost estimation model. As a final step, a crash-hotspot identification methodology comparison was implemented with the medical cost of injured persons for car crashes in Puerto Rico. With the data available (crash data and roadway information), four different high crash location performance measures were implemented and compared: 1) average crash frequency; 2) equivalent property-damage-only average crash frequency with weighting factors; 3) Jaramillo safety performance function for the identification of high-crash locations; and 4) probability of specific crash types exceeding threshold proportion using the statistical method G*. This research developed a framework for injury-only crash-cost estimation models, KABCO injury-only severity levels estimation, and identification of high crash locations of vehicle crashes with injured persons in Puerto Rico. The information presented in this study provide a better understanding of the medical costs related to crashes in Puerto Rico. It could provide information for highway safety managers to make evidence-based decisions related to highway safety improvements.
Keywords
Hotspots Identification Methods,
KABCO,
Crash Severity,
Crash Medical Cost,
Record Linkage
KABCO,
Crash Severity,
Crash Medical Cost,
Record Linkage
Usage Rights
All Rights Reserved / restricted to Campus
Persistent URL
Cite
Bianchi-Santiago, J. (2020). A framework with a cost estimation model and the identification of high vehicle crash locations with injured persons in Puerto Rico [Dissertation]. Retrieved from https://hdl.handle.net/20.500.11801/2628