Artificial intelligence applications in environmental monitoring: machine learning-based prediction of potentially toxic elements (PTEs)
Guardado en:
| Autores: | , |
|---|---|
| Formato: | artículo original |
| Estado: | Versión publicada |
| Fecha de Publicación: | 2026 |
| Descripción: | Artificial intelligence (AI), and particularly Machine Learning, has emerged as a key tool for improving environmental monitoring, enabling the development of predictive models capable of estimating pollutants from easily measurable variables. This study applied machine learning techniques, specifically Support Vector Machine (SVM), to predict the concentrations of potentially toxic elements (PTEs) in plant biomonitors and road dust from the Greater Metropolitan Area of Costa Rica. A total of 160 leaf samples from Casuarina equisetifolia and Cupressus lusitanica, along with 80 road dust samples, were collected during field campaigns in 2020 and 2021. The samples were analyzed to determine their magnetic properties and metal concentrations (Fe, Cu, Cr, Ni, Pb, V, Zn, and Cd). The SVM models, compared to simple and multiple linear regressions, showed better predictive performance for most metals, with R² values above 0.7 for Fe, Cu, Cr, V, and Zn, and lower values for Pb and Ni. C. equisetifolia exhibited the best performance in terms of fit (R²) and lower error (RMSE, MAE), surpassing C. lusitanica. In contrast, road dust showed greater variability associated with its multiple pollution sources. The results demonstrated that the integration of AI into biomagnetic monitoring facilitates the identification of complex relationships between magnetic properties and PTEs pollution, confirming its effectiveness as a fast and low-cost tool for environmental assessment. |
| País: | Portal de Revistas TEC |
| Institución: | Instituto Tecnológico de Costa Rica |
| Repositorio: | Portal de Revistas TEC |
| Lenguaje: | Español |
| OAI Identifier: | oai:ojs.pkp.sfu.ca:article/8507 |
| Acceso en línea: | https://revistas.tec.ac.cr/index.php/tec_marcha/article/view/8507 |
| Palabra clave: | Máquina de soporte de vectores (SVM) elementos potencialmente tóxicos (PTEs) propiedades magnéticas plantas polvo de carretera machine learning Support Vector Machine (SVM) PTEs magnetic properties plantss street dust |