A machine learning-driven Raman spectroscopy approach for non-invasive diagnosis of non-puerperal mastitis

Area of Science:

• Biomedical Optics
• Molecular Diagnostics
• Computational Biology

Background:

• Non-puerperal mastitis (NPM) diagnosis requires timely and accurate methods.
• Peripheral blood mononuclear cells (PBMCs) are key inflammatory mediators and potential biomarkers.
• Current detection methods for PBMCs need improvement for liquid biopsy applications.

Purpose of the Study:

• To investigate Raman spectroscopy for characterizing molecular changes in PBMCs from NPM patients.
• To develop machine learning models for predicting NPM based on PBMC analysis.
• To explore the potential of PBMC liquid biopsy for NPM diagnosis.

Main Methods:

• Collected PBMCs from NPM patients and healthy controls.
• Utilized Raman spectroscopy to analyze molecular profiles of PBMCs.
• Applied machine learning algorithms (PCA, LDA, PLSDA, SVM) for diagnostic modeling.

Main Results:

• Raman spectroscopy successfully characterized molecular alterations in PBMCs.
• Machine learning models achieved high diagnostic accuracy (AUC > 0.93).
• Demonstrated the feasibility of distinguishing NPM patients from controls using PBMC analysis.

Conclusions:

• PBMC analysis via Raman spectroscopy is a viable approach for NPM detection.
• Liquid biopsy using PBMCs, Raman spectroscopy, and machine learning shows significant diagnostic potential for NPM.
• This integrated approach offers novel opportunities for early NPM diagnosis and management.