Abstract:

This work arises in a current context where large amounts of data (Big Data) are available. However, the underlying value of these data is not easily accessible (L'Heureux et al., 2017). A possible solution to this problem, which is also seen as an opportunity to obtain indications for improving the functioning of institutions, is to appeal to knowledge extraction processes from data. These processes require, precisely, large amounts of data and allow extracting useful knowledge to leverage opportunities (Mbassegue et al., 2016). In the health area, in recent years, there has been a huge increase in information systems to support hospital activities. Despite this advance, the interoperability of digital systems remains an open issue, leading to challenges in data integration. As a result, the potential that hospital data offer, in terms of understanding and improving care, is not yet fully utilized (Johnson et al., 2016). Any issue related to health is always sensitive, given its importance to the population. It is directly related to people's well-being and, especially, to the sense of security they want to have in providing basic health care. Statistical data show that the population is increasingly aging, reinforcing the importance of the existence of a good National Health System (NHS) (Bárrios et al., 2020). However, the management and planning of emergency service are complex, causing hospitals to fail to respond within the expected time. Thus, they imply the provision of a service conditioned by resources, which are often scarce and expensive. This situation is further aggravated by the high number of patients, which can lead health professionals to make decisions under pressure (Sakellarides, 2020). Process Mining may help organizations, since it is oriented to obtain knowledge about a certain process in execution and allows to have a real model of the process behavior. It is possible to evaluate it to improve its implementation (Hendricks, 2019). In an ideal scenario, it will be possible to describe what happened, why it happened, what will happen and what can be improved on what will happen (Kurniati et al., 2018). It is intended, therefore, to analyze an emergency service and understand how Process Mining can help to provide information to managers. The idea is to help them to identify and act on existing inefficiencies, in order to design interventions that can reduce waiting time, reduce patient congestion and increase the quality of care with cost savings. The experimental scenario presented in this paper uses a dataset from the MIMIC-III database (Kurniati et al., 2018), with the information and structure necessary for the application of process discovery algorithms. For the discovery of the real model it was concluded that the Disco (Lohmann, 2012) is the simplest and most intuitive tool to use, confirming the thesis that this arises from the fact that process analysts need a tool that, above all, make Process Mining faster and easier (Lohmann, 2012). This document is organized into 5 sections. In Section 2, the main concepts that will be used throughout this paper are analyzed. Section 3 explains the data and experimental setting. Section 4 intends to present the results and the respective analyses. Finally, Section 5 concludes and sets out possible future work.