Theoretical and Natural Science · Apr 20, 2026

Predicting Alzheimer's Disease: A Machine Learning Approach Using Protein Biomarkers and Demographic Features

By Lou H

Alzheimer's disease (AD) is a dementia that primarily affects the elderly, characterized by a decline in cognitive function and daily living abilities. With the increasing global aging population, low-cost and convenient early screening methods for Alzheimer's disease can effectively promote early and regular screening among the elderly, thereby driving early intervention and treatment. This study uses clinical assessments, protein biomarkers, and demographic data from the ADNI dataset, employing random forest, XGBoost, and LightGBM models for analysis. The results show that the random forest model, optimized with GridSearchCV, performed best in both accuracy and recall. Based on this, the importance of features output by the random forest model is analyzed, and the impact of each feature on the prediction results is further explored, using the `predict_proba` function of `RandomForestClassifier`, while misclassification analysis is also performed. The results show that ABETA42, PTAU, and age are key features in early Alzheimer's disease screening, while ABETA42, PTAU, and sex play important roles in misclassification analysis. This study provides an effective machine learning method for early screening of Alzheimer's disease, which reduces screening costs while ensuring accuracy and has broad clinical application potential.

doi.org/10.54254/2753-8818/2026.32825Open accessBiomarkers

AgingResearch.News