Multimodal Early Alzheimer's Detection, a Genetic Algorithm Approach with Support Vector Machines.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Artificial Intelligence

Background:

• Alzheimer's disease (AD) is a progressive neurodegenerative disorder impacting older adults.
• Accurate diagnosis of AD and mild cognitive impairment (MCI) is crucial for timely intervention.
• Current diagnostic methods rely on biomarkers like beta-amyloid, tau protein, and brain morphology changes.

Purpose of the Study:

• To propose a novel multivariate machine learning model for enhanced Alzheimer's disease detection.
• To develop a robust biomarker by integrating diverse patient data for AD diagnosis.
• To improve the accuracy and efficiency of diagnosing AD and its prodromal MCI stage.

Main Methods:

• Utilized Alzheimer's Disease Neuroimaging Initiative (ADNI) baseline data, including 1024 clinical and neuropsychological features from 106 patients.
• Implemented data normalization and a genetic algorithm for optimal feature selection.
• Developed and validated a multivariate classification model using a support vector machine (SVM) with five-fold cross-validation.

Main Results:

• The developed multivariate model achieved an Area Under the Curve (AUC) of 87.63% during five-fold cross-validation.
• An independent blind test on 20 new patients demonstrated a perfect AUC of 100% for the final model.
• Feature selection using a genetic algorithm identified significant indicators for AD detection.

Conclusions:

• The proposed machine learning methodology effectively integrates diverse features for accurate Alzheimer's disease detection.
• The model's high performance in independent testing suggests its potential as a reliable diagnostic tool.
• This approach offers a promising avenue for early and accurate diagnosis of AD and MCI.