Cognitive and MRI trajectories for prediction of Alzheimer's disease

Area of Science:

• Neuroscience
• Medical Imaging
• Machine Learning

Background:

• Mild Cognitive Impairment (MCI) represents an early stage of Alzheimer's disease (AD).
• Early identification and intervention are critical for managing AD progression.
• Distinguishing between stable MCI and conversion to AD, and between healthy controls and conversion to MCI, is clinically significant.

Purpose of the Study:

• To predict the conversion from healthy cognition to MCI and from stable MCI to AD.
• To evaluate the added value of magnetic resonance imaging (MRI) data to cognitive tests in these predictions.
• To utilize longitudinal data and machine learning for early AD detection.

Main Methods:

• Longitudinal data from the ADNI database were analyzed.
• Mixed effects models derived features representing changes in cognitive and MRI measures.
• Ensemble machine learning models were trained to classify subgroups: healthy control (HC) vs. conversion to MCI (cMCI), and stable MCI (sMCI) vs. conversion to AD (cAD).

Main Results:

• Predictions for HC vs. cMCI improved significantly with MRI data, increasing accuracy (e.g., [Formula: see text]-score) from 60% to 77%.
• Predictions for sMCI vs. cAD showed a smaller improvement with MRI data, with [Formula: see text]-scores of 77% without and 78% with MRI features.
• Cognitive changes may appear years after AD pathology is established in the brain.

Conclusions:

• Integrating MRI features into machine learning models substantially enhances the prediction of early-stage AD conversion (MCI).
• While MRI provides significant benefits for predicting MCI conversion, its impact is less pronounced for predicting AD conversion from established MCI.
• These findings underscore the potential of multimodal data (cognitive and imaging) for early and accurate AD diagnosis.