Developed Prototype of Artificial Intelligence Techniques to Determine the Probability of Survival in Trauma Injuries

Abstract

Determining the probability or likelihood of survival in trauma injuries is important for triage, setting treatment priorities and research and management audit. The existing methods for this purpose have short comings that necessitate further development. In this study, an artificial intelligence method called fuzzy inference system (FIS) for determining the likelihood of survival in trauma injuries is being designed and evaluated. FIS is able to model complex and imprecise data in an accurate and manageable manner. The mapping between its inputs (i.e. injury information) and output (i.e. Probability of survival) is performed by a set of conditional IF-THEN rules contained in its knowledge base. The accuracy of the FIS primarily depends on the design of its knowledge base. The required knowledge base is being designed by carrying out a detailed statistical analysis of the trauma injury profiles contained in a large data base of injury cases made available to the study by a collaborating institution, Trauma and Audit Research Network (TARN).
 Currently an initial prototype of the FIS system has been developed. The aim is to finalise its design and compare its performance against the existing methods of determining the probability of survival in trauma injuries.