Implementation of Random Forest Method with Information Gain Selection and Hyperparameter Tuning for Alzheimer’s Disease Classification

Abstract

Alzheimer's disease is one of the leading causes of decreased quality of life in the elderly aged 65 years and above. One of the problems facing Alzheimer's cases is the difficulty of making an early diagnosis to prevent disease progression, as early symptoms are often mistaken for senile dementia. Using the Random Forest method with information gain feature selection and hyperparameter tuning optimization, this study aims to determine the results of optimization with feature selection and hyperparameter tuning using Random Search and Grid Search to classify Alzheimer's medical record data consisting of 32 variables, including lifestyle factors, clinical measurements, cognitive and functional assessments, as well as symptoms that indicate Alzheimer's. The results showed that applying Information Gain and parameter optimization with the Grid Search method achieved the highest accuracy among all tested experiments. Random Forest with Information Gain and Grid Search gave an accuracy of 95.57%, sensitivity of 92.93%, and specificity of 96.99%, which showed better performance than the Random Search method. This indicates that parameter optimization has a vital role in improving model performance. This research contributes to assisting paramedics in determining whether a patient has Alzheimer's disease based on the characteristics derived from the data.