Establishment and application of cardiovascular drug interaction prediction model
OBJECTIVE To predict potential cardiovascular drug interactions and their possible effect,intensity,pathway and mechanism to provide references for rational drug dosing in clinical practices.METHODS The interaction information and related drug properties of 167 cardiovascular drugs in China were collected.C5.0,CHAID and neural network algorithm were uti-lized for constructing a prediction model.RESULTS A total of 38 390 real drug interaction and 19 242 drug attribute items were collected.The association model yielded 50"drug-attribute"results,158"drug-drug"interaction items and 2"attribute-attribute"results.C5.0 model yielded 322 pharmacodynamic effects,4 180 strengths,3 406 pathways and 4 047 mechanisms of action.A total of 158"drug-drug"interaction items were verified by DDinter and Drugbank and there were 20 and 26 interaction prediction information items.The accuracy of prediction model of drug interaction effect,intensity,pathway and mechanism was 89.5％,81.0％,84.5％and 77.5％respectively.CONCLUSION Two types of models may predict the potential drug interaction,its effect,intensity,pathway and mechanism.The prediction results are reliable so that it provides rationales for laboratory studies of potential drug interactions.
drug interactioncorrelation modelC5.0 modelcardiovascular drugsdata mining
NETL
NSTL
万方数据
