Drug discovery, a pivotal aspect of pharmaceutical research, involves the identification and development of new therapeutic compounds. However, the process is hampered by challenges such as the labour- intensive nature of screening vast chemical libraries and the need to predict potential toxicity accurately. Common issues include limited labelled toxicity data, the complexity of molecular structures, and the time-consuming nature of experimental validation. This project aims to leverage deep learning for drug discovery and toxicity prediction, addressing these challenges by designing robust models capable of simultaneously predicting drug efficacy and toxicity. Through the analysis of diverse datasets, the project seeks to expedite the identification of promising drug candidates while ensuring safety. The approach involves training deep neural networks on comprehensive datasets and implementing multi-task learning strategies, to enhance the model's performance. In drug discovery and toxicity prediction, deep learning methods have proven to be powerful tools for extracting meaningful patterns and insights from complex biological data. Random Forest Regressor, Lazy Regressor and Decision Trees are commonly employed in these applications. By integrating computational intelligence with biological insights, this project strives to revolutionize pharmaceutical research and contribute to the development of safer and more effective medications.
