Solving and Explaining Scheduling problems in Digital Health using Logic Programming

The importance of Digital Health has been increasing in recent years. This is due to the development of technologies and new challenges in the healthcare sector. The world's population is increasing and becoming older, moreover, according to the World Health Organization, in the near future there will be a shortage of medical staff of several million. In this thesis, we define and propose a solution to three healthcare scheduling problems using Answer Set Programming (ASP), a declarative language used in different complex scheduling problems. Having presented these solutions, which are representative of the way in which ASP could improve and help the Hospital organization, we present another important problem that arises with the usage of AI: Explainable Artificial Intelligence (XAI). XAI is a field of AI that tries to create human-understandable solutions. In the healthcare domain, proposing a black-box model as a solution will not be enough in the future, both because the patients and the operators need to know how and why a certain solution and a certain decision was made and because the European Commission has established, with the General Data Protection Regulation, that each person has the right to ask for an explanation of the decision taken by an AI. Without developing Explainability methodologies the usage of AI based solvers will be limited, thus, both for ethical and legal reasons the implementation of explainability techniques will be crucial in all the fields in which AI could be applied and even with more urgency in the healthcare domain. To these means, we present two tools with different objectives. The first one is an explainability tool, \textsc{E-ASP}, that is used to explain the reason why a solution has been given. \textsc{E-ASP} can help users receiving a solution to understand why certain results were obtained thus allowing the usage of ASP in Safety-critical domains, such as the healthcare sector. The last tool we present is \textsc{CNL2ASP}, which is a translation tool that translates from Controlled Natural Language sentences to an ASP encoding. The tool is used to ease the usage of ASP even for non-experts users and speed up the prototyping of ASP models. Finally, another important topic in AI and Digital Health is addressed, it is the problem of Fairness. In particular, in the thesis, we will propose a solution to overcome the Fairness problem in one of the presented scheduling problems.