A Study on Machine Learning Models in Detecting Cognitive Impairments in Alzheimer's Patients Using Cerebrospinal...

Area of Science:

• Neurology
• Biomarker Research
• Machine Learning in Medicine

Background:

• Cerebrospinal fluid (CSF) biomarkers, including amyloid beta 1-42, total tau (T-tau), and phosphorylated tau (P-tau), are recognized for their role in early Alzheimer's disease (AD) detection.
• Current diagnostic approaches often combine biomarker levels with clinical dementia rating scores to differentiate AD patients from healthy controls.

Purpose of the Study:

• To evaluate the efficacy of standard machine learning classifiers in differentiating dementia patients from normal controls using CSF biomarker levels.
• To assess the diagnostic performance of various machine learning models based on CSF biomarker data.

Main Methods:

• Utilized standard machine learning classifiers: Discriminant, Logistic Regression, Tree, K-Nearest Neighbor, Support Vector Machine, and Naïve Bayes.
• Employed cerebrospinal fluid biomarker levels (amyloid beta 1-42, T-tau, P-tau) as input features for the classification models.
• Assessed model performance using accuracy and Area Under the Receiver Operating Characteristic Curve (AUC).

Main Results:

• Machine learning models demonstrated the ability to distinguish between cognitively impaired subjects and normal controls, with accuracies ranging from 64% to 69%.
• The Area Under the Curve (AUC) for the receiver operating characteristics ranged between 0.64 and 0.73, indicating moderate classification performance.
• The inclusion of two specific biomarkers, amyloid beta 1-42 and T-tau, led to a modest enhancement in classification accuracy.

Conclusions:

• Standard machine learning classifiers can differentiate dementia patients from healthy controls using CSF biomarker data with moderate accuracy.
• Amyloid beta 1-42 and T-tau levels are valuable predictors and their combined use with machine learning may improve early AD detection.
• Further research could explore more advanced machine learning techniques and larger datasets to enhance diagnostic performance for Alzheimer's disease.