Detecting Alzheimer’s Patterns Using EEG Alpha Waves and Machine Learning

1. Why is This Important?

Alzheimer’s disease is one of the fastest-growing neurodegenerative disorders in the world, affecting millions of people and placing a heavy burden on families and healthcare systems. Early detection is crucial because interventions and therapies are far more effective when the disease is still in its early stages.

Have you ever heard of electroencephalography? Well, electroencephalography (EEG) is a non-invasive method for measuring brain activity. Research has shown that many patients with Alzheimer’s often have reduced alpha-band (8–12 Hz) brain activity, reflecting disrupted neural communication. Detecting these patterns using machine learning can help identify early signs of this disease, improve multiple outcomes in healthcare, and serve as a foundation for advanced brain-computer interface (BCI) applications.

2. The Process

Step 1: Simulating EEG Data

To model Alzheimer’s vs healthy brain activity, I created a dataset of simulated alpha-wave power values. Healthy individuals were modelled with higher alpha power (~50 µV²) and Alzheimer’s patients with lower alpha power (~35 µV²). This reflects known physiological differences in the literature.

# Simulated dataset:
# 100 samples
# each sample is 1-second EEG alpha-band power
# label 0 = healthy
# label 1 = early Alzheimer's (reduced alpha power)

np.random.seed(42)

healthy_alpha = np.random.normal(loc=50, scale=5, size=50)
alz_alpha = np.random.normal(loc=35, scale=5, size=50)

X = np.concatenate([healthy_alpha, alz_alpha]).reshape(-1, 1)
y = np.array([0]*50 + [1]*50)

print("Data loaded. Shape:", X.shape)

Step 2: Training the Classifier

I used a Support Vector Machine (SVM) classifier with a linear kernel. The dataset was split into training (80%) and testing (20%) sets. The SVM learned to distinguish between healthy and Alzheimer’s alpha patterns.

# Split into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)

# Train an SVM classifier
clf = SVC(kernel='linear')
clf.fit(X_train, y_train)

# Predict on test set
y_pred = clf.predict(X_test)

# Evaluate accuracy
accuracy = accuracy_score(y_test, y_pred)
print("Model accuracy:", accuracy)

Step 3: Visualizing EEG Alpha Power Differences

To better understand how alpha-wave activity differs between healthy and Alzheimer’s-like brain patterns, I visualized the simulated EEG data using a histogram. Visualizing the data helps confirm whether alpha power is a meaningful feature for classification before directly relying on machine learning results.

The histogram below compares the distribution of alpha power values for healthy individuals and individuals showing Alzheimer’s-like patterns. Healthy samples tend to cluster around higher alpha power values more often, while Alzheimer’s samples show a noticeable shift toward lower values. This aligns with neuroscience research pointing out reduced neural synchrony in Alzheimer’s disease.

Distribution of EEG alpha power for healthy versus Alzheimer’s-like brain activity.

To generate this visualization, below is the Python code that was used:

plt.figure(figsize=(8,5))
plt.hist(X[y==0], bins=15, alpha=0.7, label="Healthy Alpha Power")
plt.hist(X[y==1], bins=15, alpha=0.7, label="Alzheimer Alpha Power")
plt.xlabel("Alpha Wave Power (µV²)")
plt.ylabel("Count")
plt.title("Healthy vs Alzheimer Alpha Wave Distribution")
plt.legend()
plt.show()

Step 4: The Prediction Tool

Finally, I created a simple prediction function to simulate how the model could classify new EEG measurements:

def predict_alz(alpha_value):
    prediction = clf.predict([[alpha_value]])[0]
    if prediction == 1:
        print(f"Alpha Power = {alpha_value} → Possible Alzheimer's pattern detected.")
    else:
        print(f"Alpha Power = {alpha_value} → Normal healthy pattern.")

# Examples
predict_alz(52)
predict_alz(30)
predict_alz(40)

This allows alpha power values to be put in and see whether they resemble healthy or Alzheimer-like brain activity.

3. Future Uses

Although this project is a simplified proof of concept, it shows how EEG-based machine learning systems could play a huge and meaningful role in future healthcare applications. By identifying patterns in brain activity, similar models could help clinicians in detecting early cognitive changes before noticeable symptoms appear.

In the future, this approach could be applied to real EEG datasets collected from patients, allowing models to learn from authentic brain signals rather than simulated values. With further validation, such systems could be integrated into brain–computer interface (BCI) technologies to continuously monitor cognitive states in a non-invasive way. This could support early screening, long-term cognitive monitoring, and personalized treatment planning for neurodegenerative diseases like Alzheimer’s.

Rather than replacing medical professionals, these tools would act as decision-support systems, helping clinicians identify subtle neural patterns that may otherwise go unnoticed.

4. Improving this Project

This project can be expanded in several meaningful ways to better reflect real-world neuroscience and clinical applications. One major improvement would be analyzing data from multiple EEG channels instead of a single alpha power value. This would allow the model to capture regional differences in brain activity across areas associated with memory and cognition.

In addition, incorporating other EEG frequency bands, such as theta, beta, and gamma waves, could provide a more comprehensive picture of neural dynamics and improve classification performance. Real EEG data would also require preprocessing steps, including artifact removal, signal filtering, and normalization, which are essential for clinical reliability.

Citations

OpenNeuro, openneuro.org/datasets/ds004504/versions/1.0.8. Accessed 17 Jan. 2026.

View of Deep Neural Network Model for Automated Detection of Alzheimer’s Disease Using EEG Signals, online-journals.org/index.php/i-joe/article/view/29867/11573. Accessed 17 Jan. 2026.

W;, Klimesch. “Eeg Alpha and Theta Oscillations Reflect Cognitive and Memory Performance: A Review and Analysis.” Brain Research. Brain Research Reviews, U.S. National Library of Medicine, pubmed.ncbi.nlm.nih.gov/10209231/. Accessed 17 Jan. 2026.

Senkaya, Yeliz, et al. “Enhancing Alzheimer’s Diagnosis with Machine Learning on EEG: A Spectral Feature-Based Comparative Analysis.” Diagnostics (Basel, Switzerland), U.S. National Library of Medicine, 29 Aug. 2025, pmc.ncbi.nlm.nih.gov/articles/PMC12428020/.

Dshmkh. “Dshmkh/Azheimer-S-Classification-Using-EEG-Signals: A Non-Invasive Diagnosis Tool for Alzheimer’s Disease Using MATLAB and Python (CNN-LSTM).” GitHub, github.com/dshmkh/Azheimer-s-Classification-using-EEG-Signals. Accessed 17 Jan. 2026.