Mapping sleep-promoting volatiles in aromatic plants with machine learning: a comprehensive survey of 2300 molecules

Abstract

Sleep disturbances affect up to one-third of the global population, yet current pharmacological therapies based on insomnia medications carry notable risks and side effects. Aromatic plants have long been valued for their capacity to ease stress and promote sleep; however, bioactive volatiles driving these benefits remain poorly understood. This study presents a comprehensive survey of 2391 volatiles across 991 aromatic plants, integrated with an ensemble machine-learning approach to identify their potential sleep-promoting activity. To evaluate the predictive accuracy of our approach, five candidate volatiles were computationally prioritized for in vivo testing. Four of these (an 80% success rate) robustly induced sleep-promoting effects, as evidenced by electroencephalogram analysis and modulation of γ-aminobutyric acid (GABA) receptor expression. In parallel, this work identified plant families such as Asteraceae, Lamiaceae, and Lauraceae as particularly enriched in high-potential volatiles and highlighted individual species—including Lavandula angustifolia and Perilla frutescens—as promising candidates for further pharmacological investigation. By combining large-scale data mining, computational prediction, and in vivo experimentation, this work first provides a comprehensive understanding of the landscape of sleep-promoting volatiles and aromatic plants and offers a reusable workflow to accelerate the discovery of bioactive compounds with potential applications in medicine, functional foods, and natural therapeutics.