Prediction of Bioactive Plant-Derived Compounds for Drug Discovery Using Machine and Deep Learning Algorithms

Abstract

Plant-derived bioactive compounds have attracted a lot of interest as drug discovery in the wake of the growing need of less harmful, inexpensive and naturally available therapeutic agents. The medicinal plants are rich in a variety of phytochemicals that have promising pharmacological effects such as anti-inflammatory effects, antimicrobial effects, anticancer effects, and antioxidant effects. But standard procedures of screening possible drug candidates are time consuming, costly and experimentally intensive. This study hypothesizes the application of a machine learning model based on Convolutional Neural Networks (CNN) that can predict bioactive plant-derived compounds that can be used in drug discovery testing. Phytochemical datasets are used in the research because the publicly available medicinal plants and chemical compound databases provide the data in terms of molecular descriptors, chemical fingerprints, and structural representations of compounds. The CNN model is developed to automatically infer complicated structural patterns and latent interactions between molecular features and biological activity, allowing to correctly classify compounds as active and inactive drug candidates. To enhance model generalization and performance, data preprocessing strategies such as normalization, feature extraction, and augmentation are used. The standard performance measures (accuracy, precision, recall, F1-score, and ROC-AUC) are used to assess the proposed model. It is also compared to traditional machine learning algorithms like the Support Vector Machine (SVM), the Random Forest (RF) and the XG Boost to prove the usefulness of CNN in predictive drug discovery. Early evidence suggests that CNN-based models have higher classification accuracy and feature learning capacity to complex phytochemical structures. This study will help speed up the discovery of plant-based drugs by combining artificial intelligence with phytochemical screening, lowering the cost of experiments and enhancing the efficiency of the process of discovering new natural therapeutic compounds. The suggested model could also facilitate the pharmaceutical research to come up with effective and sustainable plant-based medicine to different ailments.