Comparison of linear discriminant analysis and binary logistic regression analysis in predicting Alzheimer's disease

Abstract

Alzheimer's disease is a neurological disease caused by several factors affecting brain functions leading to memory loss, which in turn impacts the lives of those affected and those around them, often leading to the death of many elderly people. This study aimed to diagnose this disease and predict whether a person is affected by it. For this purpose, binary logistic regression and linear discriminant analysis were used and compared to each other to arrive at a result that clarifies which of the methods was the best and most powerful in classifying and diagnosing the disease. These methods were applied to data on reached (2149) Alzheimer's patients that were obtained from kaggle. When analyzed using the statistical program SPSS v.25, the results showed the efficiency of both methods in classification and in predicting the affiliation of group elements, as the binary logistic regression analysis method was able to correctly classify (84.8%) while it was (83.2%) for the linear discriminant analysis. As for the comparison between the two models, the results showed that the sensitivity, specificity and accuracy of the linear discriminant analysis reached (83.45, 83% and 83.2%) and (74.6%, 90.4% and 82.5%) for the binary logistic regression analysis.