Evaluating the TMR Model for Multimorbidity Decision Support Using a Community-of-Practice Based Methodology

Clinical practice guidelines are typically designed for treatment of a single disease, ignoring undesired interactions for comorbid patients. A number of methods for detecting such guideline interactions have been developed, based on computer interpretable representations of guidelines. A recently published paper by Van Woensel et al. [7] compared a number of methods for detecting and resolving interactions between multiple guidelines. The current paper contributes to this comparative corpus by applying the same functional features and evaluation dimensions to the TMR method for multimorbidity decision support. Our comparison shows that TMR allows for more complex reasoning compared to some of the methods discussed in [7]. It is one of the few that supports automated detection of adverse interactions. However, it falls short on temporal reasoning and reasoning about drug dosage. Our study also represents the first independent validation of the evaluation methodology published in [7].