SSA-classifier based screening study for Alzheimer's disease.

Area of Science:

• Biomedical Engineering
• Spectroscopy
• Machine Learning

Background:

• Alzheimer's disease (AD) is the most common neurodegenerative disorder, affecting 10% of individuals aged 65 and older.
• Early and accurate diagnosis is crucial for effective management and treatment of AD.

Purpose of the Study:

• To develop and validate a novel diagnostic framework for Alzheimer's disease (AD) screening.
• To integrate plasma attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy with advanced machine learning algorithms.
• To optimize machine learning classifier performance using a modified Sparrow Search Algorithm (GSSA).

Main Methods:

• Plasma samples were analyzed using ATR-FTIR spectroscopy.
• Four machine learning classifiers (SVM, Logistic Regression, XGBoost, LDA) were employed.
• Classifiers were optimized using the GSSA and compared against the standard Sparrow Search Algorithm (SSA) and Bayesian methods.
• Performance metrics including accuracy, sensitivity, and specificity were evaluated.

Main Results:

• GSSA-optimized classifiers significantly outperformed standard SSA and Bayesian methods.
• The GSSA-XGBoost model achieved the highest accuracy (88.51%), sensitivity (95.35%), and specificity (81.82%).
• GSSA further enhanced sensitivities to 97.67% (SVM/LDA) and specificities to 81.82% (XGBoost).

Conclusions:

• The proposed ATR-FTIR spectroscopy and GSSA-optimized machine learning framework shows significant potential as a minimally invasive screening tool for Alzheimer's disease.
• This integrated approach advances spectroscopic biomarker discovery and demonstrates the efficacy of algorithmic optimization.
• XGBoost, optimized by GSSA, provides an optimal balance for AD diagnostic performance.