Dialogue between algorithms and soil: Machine learning unravels the mystery of phthalates pollution in soil

Boyou Pan ¹, Jialin Lei ¹, Bogui Pan ²

⁰College of Life Science and Technology, Jinan University, Guangzhou 510632, China; Department of Mathematics, College of Information Science and Technology, Jinan University, Guangzhou, Guangdong 510632, China.

Journal of Hazardous Materials +

|

November 23, 2024

View abstract on PubMed

Summary

Machine learning accurately predicts soil phthalates (PAEs) accumulation, identifying hydrometeorological and soil factors as key drivers. This approach offers efficient pollutant risk assessment for agricultural sustainability and food security.

Area Of Science

Environmental Science
Agricultural Science
Data Science

Background

Soil contamination by phthalates (PAEs) poses risks to agricultural sustainability and food security.
Traditional methods for assessing PAE dynamics in soil are time-consuming and inefficient.
Developing advanced predictive models is essential for effective environmental management.

Purpose Of The Study

To develop an intelligent machine learning framework for predicting soil PAE concentrations.
To identify key factors influencing the spatial distribution and accumulation of PAEs in soil.
To provide insights into future PAE pollution trends for risk assessment and management.

Main Methods

Incorporated thirty features, including pollutant levels, agricultural inputs, soil properties, and climate data.
Developed and evaluated six machine learning models: RFR, GBRT, XGBoost, MLP, SVR, and KNN.
Utilized feature importance and non-linear effect analyses to understand PAE influencing factors.

Main Results

The Multilayer Perceptron (MLP) model achieved the highest prediction accuracy (R²=0.8637).
Hydrometeorological factors, soil moisture, and nutritional characteristics were identified as critical drivers of PAE distribution.
Significant synergistic interactions were found among environmental covariates affecting PAE levels.

Conclusions

Machine learning provides an effective and efficient approach for predicting soil PAE pollution.
Future trends indicate declining PAE levels in coastal regions and potential accumulation in inland areas.
The study offers a novel perspective for pollutant risk assessment and management in the big data era.

Keywords:

Algorithm Machine learning Multifactor interaction Phthalate esters Soil contamination