Using Data fusion for Scoring and Weighting Information in Drug-Disease Interactions Prediction

Prediction of drug-disease interactions is one of the existing fields in drug repositioning. Drug repositioning which makes use of known drugs to new indications has turned into a challenging topic in pharmaceutical science. Many current computational methods use network-based and machine learning approaches to reposition old drug for new disease. However, they often ignore features of drugs or diseases and also the importance and role of each feature. When predicting unknown drug-disease interactions, there are diverse data sources and multiple side information available to us which can help have more accurate and reliable results. This information can be collectively mined using data fusion methods and aggregation operators. In this paper, we have proposed two computational methods, namely Scored Mean Kernel Fusion (SMKF) and Weighted Mean Kernel Fusion (WMKF) which combine various data sources and side information related to drugs or diseases at two levels, drug-drug and drug-disease level, in order to predict novel drug indications. Generally, the purpose of the present study has been to investigate the effect of side information of drugs and diseases as well as data fusion in prediction of drug-disease interactions. To this purpose, the methods were validated against a well-established drug-disease gold-standard dataset. Comparisons with some existing methods revealed that our proposed method (WMKF) outperformed and is competitive in performance.

کلیدواژه ها:

data fusion ، drug-disease interaction ، kernel fusion ، support vector machine