Analyzing Recommendations Interactions in Clinical Guidelines

Analyzing interactions on combining multiple clinical guidelines

“…When comparing the above set of interactions to those already detected in our earlier work [14], the additional interactions are due to the introduction of action type hierarchies and causation beliefs, while the improvements with respect to [1] are due to the introduction of transition derivatives and contributions. In the related work where this experiment was originally proposed [6], specific rules were manually designed for detecting a number of interactions when combining the guidelines two-by-two (OA+HT, OA+DB and DB+HT).…”

“…In the given example (R4) we analyzed the (rephrased) text and we identified the following components: a positive recommendation (should) about the execution of the care action type 1 'administer thiazide', justified by a causation belief on the promotion of the transition type 'decreasing the blood pressure' with a certain probability (often). It has a high belief strength (evidence level), which means it is supported by good medical evidence.…”

“…Furthermore, the goals are simplified as 'patient wellbeing' and omitted from the representation from now on, but they could be described in more detail, for example chemotherapy may have a positive contribution to the goal of maximizing life-time but a negative contribution to improving life-quality. 1] of recommendations belonging to a guideline g ∈ CG. It associates a deontic strength of a recommendation to each pair of action a i ∈ Act (in the context of g) and the set of its contributions C g a i ∈ ℘(C g ) involving the beliefs regarding the a i .…”

“…Therefore, a family of approaches (e.g. [6][7][8][9][10]) has emerged that aims to enhance the ଝ Invited submission as extension of [1].…”

Knowledge-Driven Paper Retrieval to Support Updating of Clinical Guidelines

“…The TMR model has been developed for representing knowledge underlying CGs, aiming at supporting different guideline-related tasks. In our previous work [4,7] we addressed the task of combining several guidelines (multimorbidity) by automatically detecting interactions among recommendations. In this paper we will: (i) check the applicability of the TMR model on supporting the update task and (ii) identify improvements required to better address this task.…”