Detection and Classification of Mild Cognitive Impairment Disease in the Elderly using Deep Learning

Elderly people served nation better and public authorities are in a position to secure their tranquility and better living conditions. The future of such people has extended with mechanical progressions and study tells that the elderly populace will turn out to be twofold in the year. The major concern for elder people is that, as they get older diseases related to cognitive impairment (Alzheimer, Vascular Dementia, and Dementia) started to begin and it's quintessential to determine those early stages by healthcare specialists. As the advancements of emerging technology are revolutionizing, the usage of Deep Learning, a class of Machine learning brings a huge potential to these fields. As a result, this research offers the following steps for an effective DL model for rapid recognition of cognitive impairment (CI): a) Data was obtained on 244 subjects from two repositories: According to the Alzheimer's Disease Neuroimaging Initiative (ADNI) website, 123 entries came from ADNI, 121 entries came from AD Repository Without Borders, and 121 entries came from ADNI, b) Preprocessing were done to remove anomalies from the raw data were the selection of instances, selection of clinical scores, imputation of missing values and Data Imbalance stages are taken care, c) Feature extraction was fuzzy logic will be used for extracting certain features for the election procedure, d) Feature Selection Using Recursive Feature Elimination (RFE) and finally e) Convolutional Neural Networks for Classification (CNN). In the research, the CNN-RFE method is superior to other state-of-the-art models (accuracy of 0.96, sensitivity of 0.97, specificity of 0.88, detection rate of 0.95, TPR of 0.95, and FPR of 0.5).