A Predictive Sequence Filling Method using Protein Structures for Early Detection of Alzheimer Disease

Abstract

Alzheimer’s disease (AD) is a generic form of dementia causing memory loss and environmental responses. AD detection is pursued using the different protein structures and their intensity based on different physical behaviors. Using the time-series protein structures the detection and is eased through the proposed neural method for structural protein filling (NC-SSF). Structural differentiations are performed using the high and low intensity profiles observed. This analysis identifies the missing inputs and thereby the fillable sequences are identified. The protein biomarker determines the maximum filling requirement as per the changes observed. The neural network is trained using this sequence required under the low and high intensity variations. This process is recurrent until maximum false rate is confined through accuracy improvements. The AD progression detection is performed by estimating the intensity under different profile filling levels. The proposed method improves accuracy, sensitivity, and specificity by 8.74%, 10.29%, and 8.84% respectively. This method reduced the false rate and MMSE by 9.85% and 10.78% respectively.