Enabling data-driven decision support in healthcare

Abstract

Healthcare professionals make decisions every day. These decisions can have a profound impact on those they concern, such as a decision on how to treat a patient. Patients and their families can experience prolonged illness or death if the decision results in an adverse event; the treating healthcare organization and society as a whole will suffer the additional clinical, economic and social impact. Getting decisions right is important for safe provisioning of healthcare. This thesis discusses ways of helping healthcare professionals make better decisions through targeted use of healthcare data and knowledge. We specifically consider these three research questions: RQ1 How can health-related data and knowledge aid decision-making? RQ2 How can healthcare professionals contribute to implementation of datadriven decision-making? RQ3 How can patient trajectory data contribute to decision-making in healthcare? The five included publications cover various aspects of data- and knowledgedriven decision-making in the healthcare domain. The application scope ranges from decision support in the electronic health record to automated clinical knowledge classification. Results from the publications are discussed in light of the research questions. Finally, we consider how recent machine learning advances may affect decision support in healthcare, both in terms of possibilities and obstacles.