M. Yu. Sidorov, M. E. Gasanov, A. A. Dzeranov, L. S. Bondarenko, A. P. Kiryushina, V. A. Terekhova, G. I. Dzhardimalieva, K. A. Kydralieva, “Machine learning-enabled prediction of ecotoxicity (EC50) of diverse organic compounds via infrared spectroscopy”, Mendeleev Commun., 2024, Volume 34, Issue 6,Pages 780

Communications

Machine learning-enabled prediction of ecotoxicity (EC₅₀) of diverse organic compounds via infrared spectroscopy

M. Yu. Sidorov^a, M. E. Gasanov^b, A. A. Dzeranov^ac, L. S. Bondarenko^a, A. P. Kiryushina^d, V. A. Terekhova^e, G. I. Dzhardimalieva^ac, K. A. Kydralieva^a

^a Moscow Aviation Institute (National Research University), Moscow, Russian Federation
^b Skolkovo Institute of Science and Technology, Moscow, Russian Federation
^c Federal Research Center of Problems of Chemical Physics and Medicinal Chemistry, Russian Academy of Sciences, Chernogolovka, Moscow Region, Russian Federation
^d A.N. Severtsov Institute of Ecology and Evolution, Russian Academy of Sciences, Moscow, Russian Federation
^e Department of Soil Science, M.V. Lomonosov Moscow State University, Moscow, Russian Federation

Abstract: A new, less time-consuming and resource-intensive approach to predicting the EC₅₀ ecotoxicity index, which is crucial for assessing the impact of compounds on ecosystems, is proposed. Efficient EC₅₀ prediction based on infrared spectroscopy data and EC₅₀ values from the EcoTOX database is achieved using machine learning. The best results with an F1-score of 0.83 were obtained with the SVC and XGBoost models.

Keywords: ecotoxicology, effective concentration, EC₅₀, feature importance, infrared spectroscopy, algae, machine learning.

Language: English

DOI: 10.1016/j.mencom.2024.10.004