发明名称 PREDICTING DRUG-DRUG INTERACTIONS BASED ON CLINICAL SIDE EFFECTS
摘要 A processor-implemented method, computer program product and system are provided for predicting drug-drug interactions based on clinical side effects. The method includes constructing a drug-drug interactions training dataset that includes pharmaceutical, pharmacokinetic or pharmacodynamics drug-drug interactions from multiple data sources for each of a plurality of drugs. The method also includes constructing side effect features for each of the drugs from side effects associated with the drugs. The method further includes building, using the drug-drug interactions training dataset, a drug-drug interactions classifier that predicts adverse drug-drug interactions for drug pairs derivable from the drugs. The method additionally includes for each of the side effects, building a two-by-two table using the side effect features, and performing a Fisher's exact test using the two-by-two table to determine whether a given one of side effects is differentially shown between positive predicted drug-drug interactions and negative predicted drug-drug interactions.
申请公布号 US2015324693(A1) 申请公布日期 2015.11.12
申请号 US201414270498 申请日期 2014.05.06
申请人 INTERNATIONAL BUSINESS MACHINES CORPORATION 发明人 Hu Jianying;Sorrentino Robert K.;Wang Fei;Zhang Ping
分类号 G06N5/04;G06F19/00;G06N99/00 主分类号 G06N5/04
代理机构 代理人
主权项 1. A processor-implemented method for predicting drug-drug interactions based on clinical side effects, comprising: constructing a drug-drug interactions training dataset that includes pharmaceutical, pharmacokinetic or pharmacodynamics drug-drug interactions from multiple data sources for each of a plurality of drugs; constructing side effect features for each of the plurality of drugs from side effects associated with the plurality of drugs; building, using the drug-drug interactions training dataset, a drug-drug interactions classifier that predicts adverse drug-drug interactions for drug pairs derivable from the plurality of drugs; and for each of the side effects, building a two-by-two table using the side effect features, and performing a Fisher's exact test using the two-by-two table to determine whether a given one of side effects is differentially shown between positive predicted drug-drug interactions and negative predicted drug-drug interactions.
地址 Armonk NY US