Framework for Predicting Alzheimer’s Disease using CNN Machine Learning Classification Techniques

Abstract

The world’s population is promptly aging, and the number of persons with dementia is predictable to propagate from 35 million today to 65 million by the year 2030. Alzheimer’s Disease (AD) is the widespread origin of dementia. Dementia is an unceasing deterioration of our day to day activity like intellectual, behavioral and social aids that interrupts a human’s ability to function independently and actively. The initial signs and indications of the disease that forgetting past and present events and communications. If the disease progresses, a person with AD will have unembellished memory impairment and finally the person lose his/her ability to carry out the day to day works. In present days, no treatment is available that can predict and cure the Alzheimer’s disease in early stages. Nearly 44 million people in global living with AD. Therefore, this has become the research direction of our proposed work. In existing model, likely 18F-FDG PET brain images from the AD Neuroimaging Initiative (ADNI) were collected. But, F-FDG itself is not a conclusive imaging biomarker for AD. The former era has produced numerous tools for the early analysis of AD, but major drawbacks of existing system is that the biomarkers are expensive, less accuracy to classify the disease and high time complexity. To overcome the above drawbacks, the predictive modeling system need to be developed for accurate prediction of Alzheimer’s disease. This proposed model focuses on prediction of the disease in the early stage. A combination of ensemble learning and deep learning techniques are used to implement and develop the model. The proposed model improves classification accuracy and reduces time complexity due to Convolutional Neural Network (CNN) classification algorithm.