Analysis of the treatment of pain and anxiety in the anesthesia care in an ERCP: Process Mining application in Heath Care

Abstract

Healthcare is seen as an essential matter in developed societies, and providing patients with an outstanding quality service has been the main core objective of healthcare organizations for decades. However, the use of the very best and new techniques to provide this high quality service made the cost of the patient’s cares dramatically increase. Nonetheless, this tendency is starting to change due to, among other factors, the recent push for healthcare changes that governments and health insurance companies are introducing. This is the reason why medical institutions have lately been facing huge pressure in order to streamline their resources, so they can be able to care for more patients, attending the increasing demand for healthcare, and reduce costs while still delivering a high quality service [Anyanwu et al., 2003; Mans et al., 2009]. High amounts of data are stored in these organizations, normally for financial administration. However, until recently they were barely used to optimize the processes. This is why the medical environment is told to be “Information Rich and Knowledge Poor” [Kaur and Wasan, 2006]. A study from the Institute of Medicine (IOM) of the National Academies stated that there is a real lack of progress in the use of Information Technology (IT) to improve administrative and clinical processes. However, this scenario is starting to change with the extended use of Information Systems in modern organizations, as well as hospitals. One of the different ways hospitals can optimize their processes is by using this valuable data in order to create standardized procedures or paths via process modeling. Normally, care paths are based on the expertise of the doctors, but finding the correct model to predict the way a doctor will react to specific cases would be outstanding. We want to discover the best outcome; however, this does not mean that it has to be the most dominant. By having standardized processes, every patient will no longer be seen as a completely different and unique case. Patients with common characteristics can be cared for in a similar way. The procedures in an operation room, even though they may vary, can be seen as more uniform and process oriented than other healthcare techniques. Achieving these standardized paths is not a trivial issue. First, one will have to decide which specific techniques to apply. Healthcare processes have to deal with an extraordinary uncertainty and healthcare organizations, because of their processes, are seen as highly dynamic, complex, ad hoc, and multi-­‐disciplinary [Rebuge & Ferreira, 2012]. That is why mining techniques will be the best option. However, these techniques are usually not meant for the medical environment, they are more likely to be used in business processes, so searching for the most suitable procedure will be key. To carry out the study, two databases of the anesthesia care of ERCP (Endoscopic Retrograde Cholangio Pancreatography) will be used. The Beth Israel Deaconess Medical Centre in Boston will provide both datasets. First, we started the study with a small dataset of 33 patients, while we were waiting for the large dataset, which contained 848 patients.Many different paths can be found using process mining techniques. Usually, to make business decisions while mining, the decision is based in the cost influence, the revenue and the operational efficiency maintaining the level of care [Silver et al, 2001]. Nevertheless, in this study, the most prominent factor will be to find a standard path, so the variance should be reduced as much as possible. Searching for a standardized path does not mean that we are searching for the dominant one, but the goal will be to discover the best outcome for each situation. If a path has got a high variance, it will probably not be worth to implement it due to inefficiency and higher costs. Applying pre-­‐processing techniques such as clustering can reduce some of the variability. Other factors to take into account are the simplicity of the path, the medical correctness and the reduction of the risk for the patient. Nonetheless, the main problem we will face is to find models that are medically correct. Previous studies failed to achieve these good models while applying process mining techniques to describe process models related to healthcare. Hence, the challenge of this study is to demonstrate that these medical processes can be analyzed after process mining techniques are applied to a dataset.