A Skiing Trace Clustering Model for Injury Risk Assessment

A Skiing Trace Clustering Model for Injury Risk Assessment

Milan Dobrota (LOGIT d.o.o. Beograd, Belgrade, Serbia), Boris Delibašić (Faculty of Organizational Sciences, University of Belgrade, Belgrade, Serbia) and Pavlos Delias (Eastern Macedonia and Thrace Institute of Technology, Kavala, Greece)
Copyright: © 2016 |Pages: 13
DOI: 10.4018/IJDSST.2016010104
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

This paper investigates the relation between skiing movement activity patterns and risk of injury. The goal is to provide a framework which can be used for estimating the level of skiers' injury risks, based on skiing patterns. Data, collected from ski-lift gates in the form of process event logs is analyzed. After initial transformation of data into traces, trace vectors, and similarity matrix, using several clustering methods different skiing patterns are identified and compared. The quality of clusters is determined by how well clusters discriminate between injured and noninjured skiers. The goal was to achieve the best possible discrimination. Several experimental settings were made to achieve and suggest a good combination of algorithm parameters and cluster number. After clusters are obtained, they are categorized in three categories according to risk level. It can be concluded that the proposed method can be used to distinguish skiing patterns by risk category based on injury occurrences.
Article Preview

Introduction

Alpine skiing and snowboarding are one of the most popular recreational winter sports, with millions of sports enthusiasts participating in it (Hildebrandt, et al., 2011). Unfortunately, they are known to pose a high risk for injury because of the high speed involved (Xiang & Stallones, 2003), the mechanical leverage of skis and the effects of cold and fatigue. The most frequent reason of injury is a fall onto the slope, followed by a collision with another person (Bladen, Giddings, & Robinson, 1993; Ruedl, Kopp, Sommersacher, Woldrich, & Burtscher, 2013). Generally, many studies are showing decline in skiing injuries among recreational skiers (mostly due to improved equipment and safety concerns), from 5 injuries per 1000 skiers-days in 1970’s (Gutman, Wisbuch, & Wolf, 1976), 4 in 1980’s (Warme, Feagin, King, Lambert, & Cunningham, 1995; Bladen, Giddings, & Robinson, 1993), 3 in 1990’s, to 1.5 per 1000 skiers-days in 2000’s (Hildebrandt, et al., 2011). It is basically a decline from 5-8 to less than 2 injuries per 1000 ski-days in last decades (Ruedl, Kopp, Sommersacher, Woldrich, & Burtscher, 2013). However, they also conclude how collisions resulting in severe and often life-threatening injuries are on the rise (Hildebrandt, et al., 2011). There are number of studies investigating skiing injuries from different perspectives: separately skiers and snowboarders, categories (reasons) of accidence and mechanisms of injury (occurring due use of ski lifts, lone skier falls, crashes and collisions with objects or another person, jumps, etc.) (Ribalaygua, 2010) and impacts of environmental characteristics (time of year, time of day, location) (McBeth, Ball, Mulloy, & Kirkpatrick, 2009); environments and activities in terms of how they relate to risk-taking trends (Morrish & Groff, 2012); separately alpine (downhill) and cross-country skiing, with or without death outcome (Xiang & Stallones, 2003); demographic and injury characteristics (injury patterns by sport, demographics, skill level, etc.), reasons and types of injuries, mostly from medical perspective (Coury, et al., 2013); types of risk factors like internal (age, skill level) and external (snow, weather conditions, slope characteristics) (Ruedl, Kopp, Sommersacher, Woldrich, & Burtscher, 2013). The common purpose of such studies is finding prevention measures such as identification of risk factors and the promotion of safety behavior. However, most of the studies in this area are circumstance-related (e.g. equipment, weather conditions, bindings) and only a few studies have looked at behavior-related factors while the risk of injury is highly dependent on skiers’ risk-taking and proper ski behavior (Hildebrandt, et al., 2011). This means that level of injuries is related closely to the behavior of skiers (Gutman, Wisbuch, & Wolf, 1976) and that identifying behavioral risk factors is very important for injury prevention (Hildebrandt, et al., 2011).

This paper describes the application of trace clustering techniques in an attempt to correlate skiing pattern (habits, preferences) with a probability of getting injured. Based on skiing event log (passing the gates on ski lifts) and based on a list of injured skiers, we wish to examine whether there are skiing patterns which cause more injuries than others. To do so, several clustering models have been evaluated by how well they discriminate between injured and noninjured skiers. It is assumed that skiing pattern is related to injury level, i.e. it is expected that clustering models could discriminate well between a higher level of injured skiers and lower level of injuries in other clusters.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 10: 4 Issues (2018): 1 Released, 3 Forthcoming
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing