Intelligent Multi-Purpose Healthcare Bot Facilitating Shared Decision Making

Patient decision aids (PtDAs) have been promoted to facilitate personalized information retrieval and decision support; nonetheless, although promoted for more than 20 years, they have generally failed to gain a foothold in the general delivery of healthcare. Intelligent interactive agent technologies could address the design features necessary to facilitate support and shared-decision making. In this thesis, we develop and build a PtDA for Prostate cancer using intelligent agent technology. The proposed system, called ALAN, has a multi-layered architecture with three layers. While the first layer (User-Interface) is responsible to effectively interact with users (patients and physicians), the bottom layer (Data) handles requests regarding storing and retrieving the data. Unlike most existing bots, our core objective is to enable ALAN with learning abilities, which can evolve in the course of time and improve its behaviour with minimum distraction of the user. To this end, reinforcement learning and deep learning algorithms are employed in the main layer, i.e., Analytical Decision Making. This research is expected to have impact on delivery of personalized healthcare.