Alzheimer's Detection Using Demographic Information with a Combined Approach of Neural Network and Grid Search CV

Around the world, many are suffering from Demen-tia. It is mainly due to the neurological disorder and the onset of Alzheimer's Disease. Numerous studies have been conducted on its biomarker, Magnetic Resonance Imaging (MRI). Over the years, many algorithms, such as Neural Networks, Support Vector Machines, and various other Machine Learning algorithms, have been implemented to classify Alzheimer's Disease (AD) or Normal Control (NC) from the input MRI images. This paper focused on an integrated framework that achieved 92.26% accuracy in differentiating AD from NC using a Neural Net-work (NN), Grid Search Cross-Validation with Random Forest, and a few demographic factors related to Alzheimer's disease. The accuracies of the other three models-NN with Principal Component Analysis (85.32%), NN with t-SNE (82.57%), and Convolution Neural Networks (77.06%)-have been compared with it. Finer results than all were displayed by the suggested model. As a future extension to this model, we can install different IoT sensors on home appliances and attach them to AD patients themselves to continuously collect heterogeneous data from their behavioural patterns of daily activities and analyse them in the cloud to trigger other alarm sensors for family members or health workers when the AD patients are alone at home. IoT sensors can help confirm the sickness and protect the patient. Hence, the transformation of current healthcare facilities from hospital-centred to patient-centred is available.