Evaluation of spatiotemporal trends and predictive modeling of non-accidental trauma utilizing geographical information systems

Abstract

Background

Acts of violence can be considered random when viewed singularly, but are appreciable as patterns and clusters of an epidemic. Violence begets violence: it has been shown that people exposed to violence are more likely to harm themselves, their families, and members of the community. Our previous manuscript on this subject demonstrated distinct clusters of violent trauma, where each subtype appeared to have its own domain.

Methods

The location, date, time of day, and mechanism of injury of all non-accidental trauma patients from 1 January 2008 to 31 December 2013 were collected and analyzed. Kernel density analysis was used to identify areas of increased activity and these were compared by year. The areas identified were mapped by their latitude and longitude. The data for the year 2013 were used to determine the potential for the predictive value of the prior 5 years.

Results

Definite trends can be observed in the temporal distribution of trauma, with a higher incidence of violent trauma occurring between 6 pm and 6 am during the 6-year period. Seasonal variation of higher amounts of violent trauma is also observed from April through August. Predictive modeling did not yield significant results for the following year or for the following month using Crimestat software.

Conclusion

Discernable trends of trauma are able to be demonstrated based upon the identifiable clusters of assault but these clusters do not remain constant from year to year. Predictive modeling can assist with the identification of elevated incidents of activity, however has poor predictive value for the entirety of the year. The use of the Crimestat Time Series Forecasting Module is best served with the early indication of large changes in assault patterns. Further study and improved collaboration with the local police precincts and surrounding trauma centers is required to treat the epidemic of violence.