bridging the gap between artificial intelligence and nanotoxicity

Aim and background: nanomaterials have increased the use of nanoparticles in our daily lives, but this has also increased the threats to our health and the environment. nanotoxicology examines the toxicity of nanomaterials, which can be affected by factors such as size, configuration, and surface functioning. ai and machine learning algorithms can be used to simulate and model nano-bio interactions, gaining unique insights into biological functions.methods: regression, decision trees, support vector machines, and artificial neural networks are the primary machine learning models for predicting nanotoxicity. regression analysis is limited to linear relationships and can be influenced by outliers. decision trees can be used in quantitative structure-activity relationships to predict toxicity. support vector machines can deal with collinear descriptors, nonlinear relations, small and large datasets, and overfitted models. an artificial neural network has drawbacks such as difficulty selecting the optimal level of complexity and overfitting issues.results and discussion: quantitative structure-activity relationships are the most researched method for assessing nanomaterial-induced toxicity. it is based on physicochemical characteristics and theoretical descriptors of molecules and has the highest prediction accuracy among modern models based on calculations for physicochemical properties.conclusion: artificial intelligence is being applied to medicine, with a small amount of prospective validation done for tasks that machines could perform to assist toxicologists, regulatory bodies, and clinicians in predicting potential outcomes. there are enormous opportunities for implementing ai approaches for nanovigilance with little risk involved, which has a positive effect on healthcare.