Drug-Disease data integration for drug repurposing using deep neural networks

The coronavirus disease has led to a rush to repurpose existing drugs, although the underlying evidence base is of variable quality. Drug repurposing is becoming an increasingly attractive direction in drug discovery and development because it involves potentially shorter development timeline and lower development cost than designing a new drug. According to statistics, only ۳۰% of the drugs designed and manufactured each year are approved by the Food and Drug Administration (FDA), which proves the reasons for the necessity of using existing approved drugs for various diseases (drug-repurposing). Drug repurposing needs to consider di↵erent aspects of drugs and diseases in order to e ciently find new targets for an existing drug. In this research, we propose a model for integration of di↵erent data related to drug and disease. Then, we employ a convolutional neural network to capture similarities between data and repurpose drugs for target diseases. To this end, stratified ۵-fold cross-validation technique was used for evaluation of the methods. The results of the proposed method for the parameters of accuracy, precision, recall, and F۱-measure are ۰.۹۷, ۰.۶۹, ۰.۹۶ and ۰.۸۴ respectively. The results of comparative evaluations indicate the high performance and e ciency of the method compared to the state-of-the-art methods.

کلیدواژه ها:

drug repurposing ، deep learning ، convolutional neural networks