Biomarkers.

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Gerontology

Background:

• Alzheimer's disease (AD) is a major cause of death, impacting global health.
• T1-weighted MRI is crucial for non-invasive AD staging and risk identification.
• Demographic factors like age and gender influence AD progression and detection.

Purpose of the Study:

• To propose and evaluate a deep learning model informed by age and gender.
• To differentiate cognitively normal from cognitively impaired individuals using T1w MRI.
• To utilize the OASIS-1 dataset for model development and validation.

Main Methods:

• A six-layer fully convolutional neural network was employed.
• Age and gender data were integrated into the model.
• The OASIS-1 dataset underwent preprocessing, including skull stripping and normalization.
• A five-fold cross-validation strategy was used with weighted binary cross-entropy for class imbalance.

Main Results:

• The proposed age and gender-informed model achieved 80% average accuracy.
• The model demonstrated 96% specificity and 39% sensitivity.
• The ROC-AUC reached 0.88, outperforming a baseline model without demographic inputs.

Conclusions:

• Demographic information (age, gender) is vital for data-driven disease detection models.
• Integrating demographic factors enhances model performance for AD detection.
• This approach improves the potential for real-world clinical applications in AD diagnosis.