PyCaret machine learning library with three preprocessing steps after eLORETA source estimation predicts Alzheimer's
Yasunori Aoki1,2,3, Rei Takahashi1,2, Roberto D Pascual-Marqui4
1Department of Psychiatry, Nippon Life Hospital, Osaka, Japan.
Neuroimage. Reports
|January 21, 2026
Summary
Early Alzheimer's detection is possible using electroencephalography (EEG) and machine learning. This study shows that a linear discriminant analysis model applied to eLORETA EEG data can accurately identify Alzheimer's disease (AD) and mild cognitive impairment (MCIAD) even before symptoms appear.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Machine Learning
Background:
- Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline.
- Pathological changes in AD begin decades before clinical symptoms manifest, making early diagnosis challenging.
- Current diagnostic methods often struggle to differentiate early AD and mild cognitive impairment due to AD (MCIAD) from normal aging.
Purpose of the Study:
- To develop and validate an accurate method for early detection of Alzheimer's disease (AD) and MCIAD.
- To identify reliable biomarkers in electroencephalography (EEG) data for pre-symptomatic AD detection.
- To leverage advanced source estimation and machine learning techniques for improved diagnostic accuracy.
Main Methods:
- Utilized exact low-resolution brain electromagnetic tomography (eLORETA) for EEG source estimation.
- Applied machine learning, specifically linear discriminant analysis (LDA) via the PyCaret library, for classification.
- Employed preprocessing steps including subject-wise normalization, age-difference correction, and log-transformation on eLORETA data.
Main Results:
- The LDA model achieved 100.0% accuracy in distinguishing AD patients from healthy subjects.
- The model demonstrated high accuracy (96.4%) in identifying MCIAD patients.
- Identified specific patterns of cortical electrical activity (delta, theta, alpha, beta bands) associated with AD progression in distinct brain regions (DLPFC, PCC).
Conclusions:
- The LDA model applied to eLORETA-processed EEG data is a promising tool for early and pre-symptomatic detection of Alzheimer's disease.
- This approach can identify physiological features of AD in EEG data before the onset of clinical symptoms.
- The combination of eLORETA and PyCaret offers a significant contribution to the early diagnosis of AD, aiding in timely intervention and management.
Keywords:
Alzheimer's diseaseElectroencephalography (EEG)Exact low-resolution electromagnetic tomography (eLORETA)Linear discriminant analysisMachine learningPyCaret
