Detection of outlier behaviour amongst health/medical providers servicing TAC clients

Within the landscape of Personal Injury Compensation, building of Decision Support Tools that can be used at different stages of a client’s journey, from accident to rehabilitation, and which have various targets is important. The challenge considered in this paper is concerned with finding outliers amongst Health/Medical Providers (providers) servicing Transport Accident Commission (TAC) clients. Previous analysis by the TAC in this domain has relied upon data aggregation and clustering techniques and has proven to be restrictive in terms of providing easily interpretable and targeted results. In particular, the focus of this study is to identify outlying behaviours amongst providers rather than individual exceptional cases. We propose a new approach that enables identification of outliers on the basis of user defined characteristics.