Abstract
Adverse Drug Reaction (ADR) costs a lot of unnecessary social recourse and leads to extra pain on patients. To provide the actual information about ADR and avoid the rate of occurrence of ADR, many efforts have been done. The US Food and Drug Administration (FDA) provide a Spontaneous Reporting System Database about ADRs (AERS) which contains a lot of clinical reports from about ADRs. In this study we focus on drug-drug interaction caused ADRs. By using statistical hypothesis, characteristic of interaction-caused ADRs, and decision tree, we can find the associations between a set of drugs and symptoms, and related factors, than generate more precisely signals of interactive drug pairs and symptoms related to them. Statistical hypotheses testing can represent the association between a set of drugs and a symptom, and the characteristic of drug-drug interaction caused cases can help us evaluate the possibility of interaction. By decision tree, we can take non-drug factors into consider and help the prediction of unknown case.