Mapping cyclist activity and injury risk in a network combining smartphone GPS data and bicycle counts. (October 2015)

• Record Type: Journal Article
• Authors: Strauss, Jillian; Miranda-Moreno, Luis F.; Morency, Patrick
• Is Part Of: Accident analysis and prevention. Volume 83(2015)
• Page Start: 132
• Page End: 142
• Publication Date: 2015-10
• Subjects: Average annual daily bicycle flow -- Bayesian methods -- Cyclist safety -- Cyclist risk -- GPS smartphone application; Accidents -- Prevention -- Periodicals; Accident Prevention -- Periodicals; Accidents -- Prévention -- Périodiques; 363.106
• DOI: 10.1016/j.aap.2015.07.014
• Languages: English
• ISSNs: 0001-4575

Abstract

Highlights: Uses GPS cyclist trip data collected from a smartphone application. Obtains AADB from GPS data from extrapolation of short- and long-term count data. Bayesian prediction of injuries and risk at all intersections and segments. Maps bicycle activity and risk for the entire network of intersections and segments. Abstract: In recent years, the modal share of cycling has been growing in North American cities. With the increase of cycling, the need of bicycle infrastructure and road safety concerns have also raised. Bicycle flows are an essential component in safety analysis. The main objective of this work is to propose a methodology to estimate and map bicycle volumes and cyclist injury risk throughout the entire network of road segments and intersections on the island of Montreal, achieved by combining smartphone GPS traces and count data. In recent years, methods have been proposed to estimate average annual daily bicycle (AADB) volume and injury risk estimates at both the intersection and segment levels using bicycle counts. However, these works have been limited to small samples of locations for which count data is available. In this work, a methodology is proposed to combine short- and long-term bicycle counts with GPS data to estimate AADB volumes along segments and intersections in the entire network. As part of the validation process, correlation is observed between AADB values obtained from GPS data and AADB values from count data, with R -squared values of 0.7 for signalized intersections, 0.58 for non-signalized intersections and between 0.48 and 0.76 for segments with and without bicycle infrastructure. The methodology is also validated through the calibration of safety performance functions using both sources of AADB estimates, from counts and from GPS data. Using the validated AADB estimates, the factors associated with injury risk were identified using data from the entire population of intersections and segments throughout Montreal. Bayesian injury risk maps are then generated and the concentrations of expected injuries and risk at signalized intersections are identified. Signalized intersections, which are often located at the intersection of major arterials, witness 4 times more injuries and 2.5 times greater risk than non-signalized intersections. A similar observation can be made for arterials which not only have a higher concentration of injuries but also injury rates (risk). On average, streets with cycle tracks have a greater concentration of injuries due to greater bicycle volumes, however, and in accordance with recent works, the individual risk per cyclist is lower, justifying the benefits of cycle tracks.