A Hybrid Learning Approach for Early-Stage Prediction and Classification of Alzheimer's Disease Using Multi-Features

Alzheimer's disease, is brain disorder, and has no medicine or cure. But, at very early stage, the prediction helps to control and prevent its progression. The main outcome of this research work is, early-stage prediction and stage wise classification of Alzheimer. With the help of stage wise classification and prediction, immediate remedies can be provided and prevent the next stage of illness. Various techniques and methodologies are involved to prevent and classify this disease. But, till now, the gap in the research is, about the number of features, prediction rate and accuracy, which is not up to the mark. So, in this work, novel hybrid method called MF-RFBK method is introduced. It is based on, the boosting random forest ensemble learning, kernel density algorithm, it also uses, brief neural networks with multiple features such as Voxel-based morphometry, Cortical and subcortical volumetric and Bio-Markers features, for the prediction of Alzheimer's disease. The proposed work can improve the early-stage classification of diseases such as AD, HC, sMCI and pMCI. The boosting ensemble learning, predicts whether the given MRI image is affected by AD or not. The brief, neural network is used to classify the various stages and kernel density algorithm is used to improve accuracy of prediction. The proposed work is evaluated using, accuracy, sensitivity and specificity metrics. When the proposed work is compared with the previous works, the prediction metrics rates have increased, and it also has produced better results