Advances of machine learning-assisted small extracellular vesicles detection strategy

Area of Science:

• Biomedical Engineering
• Data Science
• Nanotechnology

Background:

• Extracellular vesicles (EVs), especially small EVs (sEVs), are vital for physiological studies and clinical applications.
• The heterogeneity of sEVs is key to distinguishing cell types and diseases.
• Conventional detection methods struggle with accurate sEV classification.

Purpose of the Study:

• To review machine learning (ML)-assisted detection strategies for sEVs.
• To highlight ML's role in cell identification and disease prediction using sEVs.
• To evaluate and compare various ML models for sEV analysis.

Main Methods:

• Application of ML algorithms like PCA, LDA, PLS-DA, XGBoost, SVM, KNN, and deep learning.
• Integration of ML with analytical techniques: surface-enhanced Raman scattering, electrochemistry, ICP-MS, and fluorescence.
• Comparative analysis of different ML-based detection strategies.

Main Results:

• ML demonstrates significant capability in overcoming limitations of conventional sEV detection.
• Various ML models have been successfully applied for sEV detection, identification, and classification.
• The review compares the performance, advantages, and limitations of different ML approaches.

Conclusions:

• ML-powered strategies offer promising avenues for sEV analysis, cell identification, and disease prediction.
• Further research into ML models and their integration with detection techniques is warranted.
• Addressing the merits and limitations of current approaches will guide future directions in sEV research.