NeuroSplit: Differentiating ADHD and Early Alzheimer’s Using EEG Biomarkers
Faculty Mentor
Meenalosini Vimal Cruz
Location
Savannah Ballroom
Type of Research
Completed
Session Format
Poster Presentation
College
Allen E. Paulson College of Engineering & Computing
Department
Department of Information Technology
Abstract
Electroencephalography (EEG) is a powerful and noninvasive technique of neurophysiological recording and biomarkers of cognitive impairment. Attention Deficit Hyperactivity Disorder (ADHD) and Alzheimer's Disease (AD) are two extreme poles of the neurological impairment: one developmental and one degenerative, but they have similar behavioral and cognitive traits that make it more difficult to distinguish them. The paper provides a comparative model of distinguishing the EEG patterns of ADHD and Early Alzheimer's Disease based on spectral analysis and feature learning using deep learning. The methodology includes strict preprocessing, such as band-pass filtering, segmentation, and normalization, estimation of power spectral density, and calculation of band-power in the delta, theta, alpha, beta, and gamma frequencies. These spectral features were modelled as a Multilayer Perceptron (MLP-Net) architecture that consisted of fully connected layers, ReLU activation, batch normalization, and dropout regularization. The model was shown to be very effective in capturing complex spectral-spatial correlations and also reached a classification accuracy of 99.92, which was supported by high levels of precision and recall scores. The results reveal theta enhancement in ADHD and alpha suppression in Alzheimer's EEG, which is in line with the neurophysiological findings. In general, the framework has a high potential of being a scalable and interpretable basis of future clinical systems developed to detect early neurological disorders.
Program Description
.
Start Date
4-21-2026 1:30 PM
End Date
4-21-2026 3:30 PM
Recommended Citation
Kumar NavaneethaKrishnan, Nithish, "NeuroSplit: Differentiating ADHD and Early Alzheimer’s Using EEG Biomarkers" (2026). GS4 Student Scholars Symposium. 77.
https://digitalcommons.georgiasouthern.edu/research_symposium/2026A/2026A/77
NeuroSplit: Differentiating ADHD and Early Alzheimer’s Using EEG Biomarkers
Savannah Ballroom
Electroencephalography (EEG) is a powerful and noninvasive technique of neurophysiological recording and biomarkers of cognitive impairment. Attention Deficit Hyperactivity Disorder (ADHD) and Alzheimer's Disease (AD) are two extreme poles of the neurological impairment: one developmental and one degenerative, but they have similar behavioral and cognitive traits that make it more difficult to distinguish them. The paper provides a comparative model of distinguishing the EEG patterns of ADHD and Early Alzheimer's Disease based on spectral analysis and feature learning using deep learning. The methodology includes strict preprocessing, such as band-pass filtering, segmentation, and normalization, estimation of power spectral density, and calculation of band-power in the delta, theta, alpha, beta, and gamma frequencies. These spectral features were modelled as a Multilayer Perceptron (MLP-Net) architecture that consisted of fully connected layers, ReLU activation, batch normalization, and dropout regularization. The model was shown to be very effective in capturing complex spectral-spatial correlations and also reached a classification accuracy of 99.92, which was supported by high levels of precision and recall scores. The results reveal theta enhancement in ADHD and alpha suppression in Alzheimer's EEG, which is in line with the neurophysiological findings. In general, the framework has a high potential of being a scalable and interpretable basis of future clinical systems developed to detect early neurological disorders.
