Tackling Data Related Challenges in Healthcare Process Mining using Visual Analytics

Abstract

Data-science approaches such as Visual analytics tend to be process blind whereas process-science approaches such as process mining tend to be model-driven without considering the "evidence" hidden in the data. Use of either approach separately faces limitations in analysis of healthcare data. Visual analytics allows humans to exploit their perceptual and cognitive capabilities in processing data, while process mining represents the data in terms of activities and resources thereby giving a complete process picture. We use a literature survey on both Visual analytics and process mining in the healthcare environments, to discover strengths that can help solve open problems in healthcare data when using process mining. We present a visual analytics approach in solving data challenges in healthcare process mining. Historical data (event logs) obtained from organizational archives are used to generate accurate and evidence based activity sequences that are manipulated and analyzed to answer questions that could not be tackled by process mining. The approach can help hospital management and clinicians among others, audit their business processes in addition to providing important operational information. Other beneficiaries include those organizations interested in forensic information regarding individuals and groups of patients.