Analysis of Clinical Discussions Based on Argumentation Schemes

Clinical discussions usually taking place in healthcare structures allow medical specialists to focus on critical cases, debate about different diagnostic hypotheses, identify therapeutic protocols, or choose among alternative treatments. This paper presents an argumentation-based approach to the analysis of clinical discussions, with the aim of providing a multidisciplinary medical team with a support tool that may help discover whether clinical discussions are affected by any weak points, such as contradicting conclusions, invalid reasoning steps, hidden assumptions, or missing evidences. To this end, we have adopted an approach based on argumentation schemes, which provide an intuitive yet well structured representation of general reasoning patterns. Argumentation schemes include one or more premises, a conclusion, and a set of critical questions that challenge the validity of the relation between premises and conclusion. We exploit argumentation schemes to interpret the assertions made by the participants in a meeting and to generate a graph of arguments connected through edges that represent support or attack relations existing among them. The resulting graph is then used to carry out a logical analysis of the discussion, highlighting, for instance, conflicting opinions or suggesting the need for gathering additional information. To show the potential of our approach, we have developed a sample case based on a clinical discussion taken from literature. After having identified a set of argumentation schemes appropriate for the medical domain considered, the case has been analyzed and a detailed logical analysis has been carried out.