Edinburgh Napier University

  Self-management of type 1 diabetes is a challenging and complex task due the constant need for self monitoring and the diverse range of factors to consider in order to effectively regulate blood glucose levels. Recommender systems have been demonstrated to be effective for supporting patient self-management of type 1 diabetes by providing recommendations for insulin doses. Recent studies have expanded on this approach by incorporating case based reasoning within existing recommender systems for type 1 diabetes, to provide a more flexible and personalised approach to making recommendations. However, recommendations made by such systems may be ignored, even when users consider the system’s performance to be good. To address this, we propose a complimentary approach to increase the transparency of such systems through the provision of explanatory summaries that expose the reasoning process for making the recommendation. Greater transparency may increase recommendation acceptance rates and improve users’ trust and acceptability of these systems.