Decision support recommendation optimization

A non-transitory computer-readable media having computer-executable instructions embodied thereon that when executed provide a method for providing a decision support recommendation service, the method comprising:

• a) accumulating medical transaction data associated with a user including patient data for past events;
• b) accumulating operational data indicating recommendations previously presented to a user and an action of the user associated with the recommendation;
• c) determining a frequent itemset associating a recommendation and a user acceptance of said recommendation based on said action of a user;
• d) generating a set of association rules for generating recommendations based on said operational data, said medical transaction data and said frequent itemset;
• e) processing the set of association rules for likelihood of predictive power;
• f) receiving an event and a recommendation context;
• g) identifying a plurality of candidate recommendations predicted by the association rules to match the recommendation context;
• h) determining to refrain from issuing a qualifying candidate recommendation based on the likelihood of predictive power with regard to predicting recommendation uptake.

A method for providing recommendations to a provider based on transaction data for a plurality of patients, the method comprising:

• a) assembling transaction data for a plurality of patients;
• b) determining a patient context comprising serial medical data;
• c) generating a recommendation to a provider regarding the patient context;
• d) collecting information about provider acceptance of said recommendation;
• e) accumulating an ensemble of recommendation information;
• f) identifying a pattern that is predictive of recommendation acceptance over said ensemble;
• and
• g) determining to refrain from emitting a recommendation based on said pattern, the determining to refrain from emitting a recommendation comprising filtering the ensemble based on said pattern.