A Machine Learning Perspective for Early Alzheimer’s Diagnosis

Alzheimer’s disease (AD) is one of the most common types of dementia which belongs to the category of neurodegenerative diseases (NDDs). The AD is associated with the symptoms of losing memory, language, losing control of the nervous system. The disease has a major impact on people who are in their mid-60s or later, these people tend to have Late-Onset Alzheimer’s Disease (LOAD) and those who develop the disease before 65 are considered to have Early-Onset Alzheimer’s Disease (EOAD). This research mainly focuses on detecting AD at its early stages by identifying the biomarker that is effective to bring out the pathogenesis of AD. The biomarker that is effective so far is the beta-amyloid 42 and tau. These proteins tend to deposit in the neurological paths of the brain that is accountable for memory and language. These components cause the tangling of neurofibrils and deposition of proteins in the narrow paths connecting to brain in an abnormal level which leads to the onset of AD. According to recent research, the immune system and the central nervous system (CNS) are intricately linked. An imbalance in both systems can lead to neuroinflammation and Alzheimer’s disease (AD). One theory about why amyloid builds up in the brain is due to protein malfunction, but what causes such malfunction is still unknown. This led to identifying biomarkers assuming that they might be responsible for the onset of the disease. The main objective of this study is to detect AD at its early stages using machine learning algorithms. The proposed methods are SVM, logistic regression, decision tree, and random forest out of which random forest yielded a maximum accuracy of 86%.