Domain-driven KDD for mining functionally novel rules and linking disjoint medical hypotheses

Introduction: An important quality of association rules is novelty. However, evaluating rule novelty is AI-hard and has been a serious challenge for most data mining systems. 

Objective: In this paper, we introduce functional novelty, a new non-pairwise approach to evaluating rule novelty. A functionally novel rule is interesting as it suggests previously unknown relations between user hypotheses. 

Methods: We developed a novel domain-driven KDD framework for discovering functionally novel association rules. Association rules were mined from cardiovascular data sets. At post-processing, domain knowledge-compliant rules were discovered by applying semantic-based filtering based on UMLS ontology. Their knowledge compliance scores were computed against medical knowledge in Pubmed literature. A cardiologist explored possible relationships between several pairs of unknown hypotheses. The functional novelty of each rule was computed based on its likelihood to mediate these relationships. 

Results: Highly interesting rules were successfully discovered. For instance, common rules such as diabetes mellitus⇔coronary arteriosclerosis was functionally novel as it mediated a rare association between von Willebrand factor and intracardiac thrombus. 

Conclusion: The proposed post-mining domain-driven rule evaluation technique and measures proved to be useful for estimating candidate functionally novel rules with the results validated by a cardiologist.