Decoding myasthenia gravis: advanced diagnosis with infrared spectroscopy and machine learning

Area of Science:

• Biomedical Spectroscopy
• Neurological Disorders
• Machine Learning in Diagnostics

Background:

• Myasthenia Gravis (MG) is a rare neurological disorder with unclear mechanisms and diagnostic challenges.
• Current diagnostic tests for MG are time-consuming, expensive, and can yield negative results.
• There is a critical need for rapid, cost-effective methods for early and accurate MG diagnosis.

Purpose of the Study:

• To identify spectral biomarkers of Myasthenia Gravis (MG) in blood serum using infrared spectroscopy.
• To develop a rapid diagnostic approach for MG by coupling infrared spectroscopy with multivariate analysis.
• To evaluate the diagnostic performance of this novel method for MG.

Main Methods:

• Infrared spectroscopy was employed to analyze blood serum samples from MG patients.
• Multivariate analysis techniques, including Principal Component Analysis (PCA) and Support Vector Machine (SVM), were utilized.
• Spectral data were analyzed to identify MG-induced changes and build a diagnostic model.

Main Results:

• Significant alterations in lipid peroxidation, lipid, protein, and DNA concentrations were observed in MG patients.
• Changes in protein phosphorylation and structural dynamics, along with specific ratios (PO2-/protein, PO2-/lipid), were identified as biomarkers.
• The combined infrared spectroscopy and multivariate analysis approach achieved 100% accuracy, sensitivity, and specificity in diagnosing MG.

Conclusions:

• FTIR spectroscopy combined with machine learning provides a rapid, low-cost, and highly sensitive method for MG diagnosis.
• Identified spectral parameters serve as potential biomarkers for MG diagnosis and therapeutic monitoring.
• This technology shows promise for clinical translation, improving early detection and management of Myasthenia Gravis.