ALZHEIMER'S DISEASE DETECTION

Personal project | Machine learning | 2026

A machine learning project exploring Alzheimer's disease classification using MRI volumetric measurements, cognitive assessment scores and clinical diagnosis data. Three datasets were merged on the patient identifier into a single table, and the project follows an iterative workflow in which each notebook improves on the previous one, covering preprocessing, feature selection, model training, evaluation and optimisation.

Key contributions:

Merged MRI, cognitive score and clinical datasets into one modelling table and handled incorrect data types and missing values.

Trained a baseline Random Forest on all selected features, which reached 100% training accuracy but only 50% on the test set, a clear case of overfitting.

Improved the preprocessing pipeline and reduced the feature space in a second notebook, raising test accuracy from 50% to 80.6%.

Tuned the Random Forest hyperparameters and documented future directions such as cross-validation, recursive feature elimination and PCA.

Collaborators: Gabriela Słomiany

Software Used: Jupyter Notebook, pandas, NumPy, scikit-learn, Seaborn, Matplotlib

Programming Languages: Python

Repository: github.com/imgabrielas/Alzheimer-Detection

Project Date: July 2026