Detecting Alzheimer's Disease (Gender-Based) Using Different Machine Learning Approaches

In the current scenario, Alzheimer's Disease constitutes 60-70% of neurodegenerative issues and affects 55 million people across the globe. As this disease progresses, the symptoms become worse and almost unmanageable. Thus, the early diagnosis of Alzheimer's Disease during the onset of mild symptoms, would be highly beneficial in tackling the challenges posed by this disease. Therefore, the work performs a comparative analysis (gender-based) of various methods including Support Vector Machine, k-Nearest Neighbours, Logistic Regression, Random Forest, Decision Trees, Naïve Bayes, and Modified k-Nearest Neighbours to diagnose Alzheimer's disease. For this, the Open Access Series of Imaging Studies (OASIS) dataset is used. The experimental analysis revealed that the Modified k-Nearest Neighbours is 99.12%, 99.49%, and 90.32% accurate in identifying Alzheimer's irrespective of gender, in male-gender, and female-gender subjects respectively. Additionally, other renowned classifiers such as Random Forest and cubic degree polynomial kernel Support Vector Machine are 98.96% accurate in detecting Alzheimer's irrespective of gender. The same classifiers are 97.62% accurate in diagnosing male Alzheimer's patients. Likewise, for distinguishing female Demented subjects from Non-demented cohorts, Random Forest and conventional k-Nearest Neighbours (at k=5) are 88.71% accurate. With such results emphasizing gender parameters, modified k-Nearest Neighbours is found more effective in diagnosing Alzheimer's Disease than other renowned classifiers.