Application And Limitations of Artificial Intelligence in Natural Product Discovery

Abstract

In pharmaceutical sector, AI is significantly influential as the younger generation shifts towards natural remedies, appreciating their safety and effectiveness with minimal side effects. However, isolation of pure and specific phytoconstituents from a wide array of options is challenging. Artificial intelligence is utilized in several capacities, including the structural elucidation of chemical compounds, predicting the bioactivity of targeted substances, designing novel drugs with unique structures, and assessing the absorption, distribution, metabolism, efficacy, and toxicity of key compounds. Although it reduces human errors, it encounters difficulties such as managing the complexities of data, variability, quality, and biases. Issues related to the model, including overfitting and underfitting of datasets, disrupt the training of models on these datasets. Even the biological and experimental research, challenges arise from incomplete gene annotations and the failure to accurately replicate in-vitro and in-vivo models. This review article discusses the challenges faced by artificial intelligence during natural product drug discovery.

Keywords