RTLS-based Process Mining: Towards an Automatic Process Diagnosis in Healthcare
Résumé
Log files constitute the main data source required to be able to use a Process Mining tool. As soon as an information system enables to record events corresponding to activity changes, it is rather simple and rapid to completely and automatically model a process. In numerous fields, information systems do not record events enough detailed. Therefore the model obtained with Process Mining will not be accurate and detailed enough for further analysis. In this article, we are interested in modeling with Process Mining patient pathways in an external consulting service of a hospital center. Apart from recording patients at the front desk and when they are leaving, the information system does not collect enough events in order to manage to model in details the different patient pathways. As a result, the model must be realised with successive observations, interviews and manual collects. This work often represents a significant workload without ensuring data quality and representativeness. To resolve this issue we suggest using a Real Time Location System ( RTLS) that enables to automatically record events according to patient locations in the service. The log file obtained can contain the pathway tracks followed by the patients with enough details to precisely rebuild the process with Process Mining. This article is intended for users, for diagnosis experts who will be able to realise an accurate diagnosis and then propose improvements. This article deals with our on-going work through a real case study.