Development of computerized guidelines for the management of chronic diseases allowing to position any patient within recommended therapeutic strategies.

Chronic diseases are complex to manage. One reason comes from the difficulty to synchronize a patient's therapeutic history with the guideline-based sequence of treatments. We propose to represent guideline knowledge as a two-level decision tree, a clinical level describing theoretical clinical situations and a therapeutic level formalizing the different steps of corresponding recommended therapeutic strategies. Guideline-based strategies are first represented as bidimensional matrices structured in lines of therapy and levels of therapeutic intention. A revised version introducing levels of therapeutic combination is then developed. The therapeutic level is operationalized for any patient therapeutic history to provide the next best step of treatment. Evaluated on actual patient records, our system proved to impact physicians' decisions in 78% of the cases and led to a significant improvement of their compliance with recommendations.