Title : Temporal distribution and forecasting recreational sport injuries
Quadbikes are all-terrain four-wheeled vehicles used extensively as desert recreational vehicles in Dubai, UAE. We tried to assess the trend and temporal distribution of quadbike injuries based on Electronic Patient Care Registration (EPCR) reports of Dubai Ambulance [Dubai Corporation for Ambulatory Services] from January 1, 2017, to March 1, 2021. We would like to share our attempt to forecast short term injury trend using time series analysis on such a small dataset. We also tried to identify the temporal distribution of quadbike injuries in this unique desert ecosystem, to identify the population at risk and factors influencing injuries.
Patient case history was the source for various variables like time of incident, injury outcome, in addition demographic variables like age, gender and nationality. Temporal variables like time of injury and date were used to arrive at the binary outcome variables like nighttime injury and winter injury respectively. IBM SPSS Statistical Package Version. 28.0 was used for descriptive, bivariate and regression analysis to identify demographic factors influencing the outcome variables.
R software version 4.2.1 (“forecast” package) helped with the time series analysis. Seasonal and Trend decomposition (STL) using Loess decomposition, split the monthly time series data into trend, seasonal and noise components. We fitted the ETS (Error, Trend and Seasonal components) state-space model to the Dubai quadbike injury data using the ets() function. Best fit model was selected using Akaike Information Criterion, corrected for small sample bias (AICc).
Temporal distribution and time series decomposition showed marked seasonality where injuries peaked in winter. Similarly, injuries peaked around late afternoon and evening during the hours of a day. Nationality had a strong influence on nighttime injuries and winter injuries. The downward trend in injury occurrence from 2017 to 2021 also includes the low tourist seasons observed during the winter lockdowns in Dubai followed by a rise in the post lockdown period.
Temporal distribution helps identify the most optimal periods for injury prevention interventions. However, time trend and forecast can be challenging in small datasets with disruptions like the COVID lockdown.