Machine learning drugs side effects prediction

Abstract

Adverse drug reactions can be the cause of hospitalization, increased morbidity and mortality, withdrawal of drugs from the market and consequently increased costs of the healthcare system. Current methods for predicting and assessing potential side effects are challenging in terms of costs and efficiency. Machine learning could be implemented for predicting the side effects of drugs. Therefore, we present machine learning classifier for predicting drugs side effects using different supervised learning models on a dataset consisted of chemical, biological and phenotypic features. Compared to other machine learning models for prediction of side effects of drugs, our model has similar and comparable performance. Machine learning probably wouldn't be able to predict all side effects, but it could help scientists to notice potential problems early and develop safer drugs in the future.