|Title||Discovering Drug-Drug Interactions Using Association Rule Mining from Electronic Health Records|
|Authors||Nazyrova, N., Chahed, S., Dwek, M., Getting, S. and Chaussalet, T.|
In this paper, we propose utilising Electronic Health Records (EHR) to discover previously unknown drug-drug interactions (DDI) that may result in high rates of hospital readmissions. We used association rule mining and categorised drug combinations as high or low risk based on the adverse events they caused. We demonstrate that the drug combinations in the high-risk group contain significantly more drug-drug interactions than those in the low-risk group. This approach is efficient for discovering potential drug interactions that lead to negative outcomes, thus should be given priority and evaluated in clinical trials. In fact, severe drug interactions can have life-threatening consequences and result in adverse clinical outcomes. Our findings were achieved using a new association rule metric, which better accounts for the adverse drug events caused by DDI.
|Keywords||drug-drug interactions, association rule mining, adverse drug events, polypharmacy, hospital readmission|
|Conference||The 17th International Conference on Innovations in Intelligent Systems and Applications|
|Accepted author manuscript|