Machine Learning for Alzheimer's Disease Progression Under Extreme Class Imbalance

A 2026 study by Patrick O. Akinwumi analyzes 3,240 participants from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) to predict diagnostic worsening within 24 months. It uses baseline demographic, clinical, and cognitive variables to train XGBoost and logistic regression models under extreme class imbalance. The dataset includes model performance metrics and SHAP analyses for feature interpretation.

Use Cases

• Benchmarking predictive models for rare clinical events based on the described extreme class imbalance scenario.
• Evaluating feature importance for Alzheimer's disease progression based on SHAP analyses of baseline cognitive and functional measures.
• Comparing imputation strategies for clinical data based on the study's evaluation of KNN and MICE methods.
• Assessing model calibration and precision-recall trade-offs in medical diagnostics based on the reported AUROC and AUPRC metrics.

Strengths

• Includes data from 3,240 participants sourced from the Alzheimer’s Disease Neuroimaging Initiative.
• Evaluates multiple machine learning models (XGBoost, logistic regression) and imputation strategies, providing comparative performance metrics.
• Employs interpretable ML methods with SHAP analysis to quantify feature contributions.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is very small at 15.1 KB, indicating limited scope, likely containing summary results rather than raw participant data.

Provenance

Source

Alzheimer’s Disease Neuroimaging Initiative (ADNI)

Collection Method

Analysis of participant data using stratified k-fold cross-validation and longitudinal mixed-effects modeling.

Time Range

Study period includes 24-month follow-up data.

Freshness

Last updated 2026-05-25 05:31:11; freshness should be verified.