A Novel Multi-Classification Approach for Drugs Based on ANN and Other ML Algorithms with Hyper Parameter Tuning

In the realm of pharmaceutical research and development, accurate classification of drugs is of paramount importance for effective therapeutic interventions. This paper proposes a novel approach to drug classification utilizing a combination of Machine Learning (ML) and Artificial Neural Networks (ANN) algorithms. By harnessing the power of ANN's ability to learn complex patterns and ML's versatility in handling diverse datasets, our approach aims to enhance the accuracy and efficiency of drug classification tasks. Furthermore, we incorporate hyper parameter tuning techniques to optimize model performance, thereby maximizing predictive capabilities. Through comprehensive experimentation and evaluation on real-world drug datasets, our proposed methodology demonstrates promising results, showcasing superior classification accuracy compared to traditional methods. This research advances drug discovery and development by providing a robust framework for accurate drug classification in biomedical and pharmaceutical fields. In this study we have applied ten ML technique: Logistic Regression, Decision tree, Random forest, Catboost classifier, Adaboost classifier, Ensemble classifier, Gradient boosting classifier, LGBM classifier, Soft voting classifier, ANN with hyper parameter tuning and Support vector machine on the drugs classification dataset. The evaluation of classification outcomes was conducted both with and without the implementation of hyper parameter tuning. Notably, upon applying hyper parameter tuning, the ANN classifier emerged as the top performer, surpassing all other classifiers. RFC and DTC have also attained good accuracy of 96%. Impressively, ANN attained an accuracy rate of 99.29%, showcasing its superior performance in the dataset.