National Institute of Informatics, Tokyo, Japan
16 July 2018
Auditório DI-A2, 11h00
Multimorbidity is defined as the presence of two or more chronic medical conditions in an individual, presenting problems in care, particularly when the number of existing conditions is high and there are treatment conflicts. It poses challenges, namely in the use of multiple drugs to treat all existing conditions (drug-drug interactions) and the effects their combination may have on the patient's body or by influencing the evolution of other health conditions (drug-disease interactions). The formalization of multimorbidity decisions serves two purposes: to support the stakeholders in choosing which treatment to apply and to identify the reasons behind decisions. We investigate the use of computational argumentation to both analyse and generate decisions in multimorbity about consistent recommendations, according to the different goals of stakeholders. Decision-making in this setting carries a complexity related with the multiple variables involved. These variables reflect the concomitant health conditions that should be considered when defining a proper therapy. However, current Clinical Decision Support Systems (CDSSs) are not equipped to deal with such a situation. They do not go beyond the straightforward application of the rules that build their knowledge base and simple interpretation of Computer-Interpretable Guidelines (CIGs). We provide a computational argumentation system equipped with goal seeking mechanisms to combine independently generated recommendations, then identify and discuss its advantages over multiple-criteria decision analysis in this particular setting.