Integrating untargeted metabolomics and deep learning approaches to identify specific metabolic signatures and new mechanisms in unstable plaques

1

The study identified 98 metabolites significantly associated with unstable carotid plaques, which are critical for predicting stroke risk.
2

Machine learning algorithms, including RF, SVM, LASSO, and LR, were utilized to analyze metabolic biomarkers linked to unstable carotid plaques.
3

Metabolic pathways such as cGMP-PKG signaling and glucagon signaling were found to be significantly associated with unstable carotid plaques.
4

The research highlights the importance of metabolomics in characterizing unstable plaques and identifying potential biomarkers for stroke risk.
5

This study emphasizes the need for early detection of unstable carotid plaques to improve stroke prevention strategies.

Frontiers In Cardiovascular Medicine

by Jia-Qi Ma, Lu Wang, Xiao-Peng Qu, Yue Zhang, Li-Jia Song, Guo-Dong Gao, Chao Wang, Long-Long Zheng, Qi-Xing Fang, Yan Qu, Liang-Liang Shen, Bei Liu
May 12, 2026

1