Information usefulness as a strategy for action selection in health dialogues

The design of effective strategies for managing information within the healthcare domain presents more complex challenges with respect to the classic conversation paradigm. One of such challenges is the capability of an agent of adapting its behavior based on the context in which it is deployed. For example, while some agents have to be able to acquire enough knowledge to reduce as much as possible its uncertainty on the situation (e.g. diagnosis), an agent that deals with risky situations (e.g. asthma self-management) has to be able to reduce as much as possible the number of interactions with the patient to quickly detect those situations. In this paper, we propose a framework that supports the development of intelligent conversational agents within the healthcare domain. We introduce our framework through the use of a running example, showing how an agent acts in the conversation based on its background knowledge and on the information that it acquires from the patient. Then, we describe the framework by providing the definition and the rationale of each component. Finally, we evaluated the metric we proposed for selecting the next action with respect to a set of baselines.