Application of Modern Data Analysis Methods to Cluster the Clinical Pathways in Urban Medical Facilities

Abstract

Patient flow modeling in healthcare plays a large role in understanding the operation of the system and its characteristics. Besides, modeling techniques can significantly improve the effectiveness of the medical facilities. The existing level of automation in these facilities enables the accumulation of large amounts of various data. Therefore, the collected data might be considered as the resource of new valuable knowledge. A novel approach to automatically identify the groups of similar clinical pathways based on event hospital data is presented in the paper. More specifically, the approach summarizes the most frequent pathways by implementing hard and soft clustering algorithms in order to describe the behavior patterns. The obtained clusters of clinical pathways serve as a starting point for the development of a personalized approach in modelling the heterogeneous patient flow in urban medical facilities. The results indicate the suitability of multidimensional time series clustering and Additive Regularization of Topic Models (ARTM) for the clinical event data.