Related Experiment Video
Updated: Feb 28, 2026
09:10
Multimodal Analysis of Microplastics in Drinking Water using a Silicon Nanomembrane Analysis Pipeline
Published on: June 13, 2025
1.5K
Metabolomics‑driven, data‑augmented machine learning for predicting toxicity of microplastic mixtures
Beilei Yuan1, Chengzhi Liu2, Shuang Chen1
1College of Safety Science and Engineering, Nanjing Tech University, Nanjing, Jiangsu 210009, China.
Ecotoxicology and Environmental Safety
|February 25, 2026
Summary
Predicting microplastic (MP) toxicity is challenging due to complex mixtures. A metabolomics-driven machine learning model effectively predicted MP cytotoxicity, offering insights into cellular energy metabolism reprogramming.
Area of Science:
- Environmental Science
- Toxicology
- Computational Chemistry
Background:
- Microplastics (MPs) exist as complex mixtures in the environment, hindering individual toxicity assessments.
- Developing rapid and effective methods to evaluate the toxicity of MP mixtures is crucial for risk assessment.
Purpose of the Study:
- To develop predictive models for evaluating the toxicity of microplastic mixtures.
- To compare the performance of quantitative structure-activity relationship (QSAR), quantitative bioactivity relationship (QBAR), and quantitative structure-bioactivity relationship (QSBAR) models.
Main Methods:
- Explored three model frameworks: QSAR, QBAR, and QSBAR.
- Employed six machine learning algorithms with data augmentation strategies.
- Utilized metabolomics data to screen biodescriptors for the QBAR model.
Main Results:
- The QBAR-based eXtreme Gradient Boosting (XGB-qbar) model demonstrated superior performance (R²test = 0.8923).
- Key biological descriptors influencing toxicity were identified.
- Metabolomics analysis revealed MP mixture exposure may reprogram cellular energy metabolism.
Conclusions:
- A metabolomics-driven, data-augmented machine learning approach efficiently predicts microplastic toxicity in complex mixtures.
- This approach provides mechanistic insights and a feasible pathway for environmental risk assessment.
