Alzheimer’s Disease Evaluator

Objective: The onset of Alzheimer’s disease(AD)is insidious and imperceptible.Therefore,the early detection and treatment for the disease is of great significance to the patient’s family and the society.At present,major screening methods for the disease require the participation of professional doctors,which make it difficult to be popularized among families and communities.The main purpose of this paper is to design an evaluation instrument for AD and to provide a feasible solution for automatic screening of AD.Methods: The evaluation instrument includes test software and evaluation algorithm.The test software was used to collect indicators related to eye-movement and cognitive psychology from the subjects,and the evaluation algorithm could find out the AD patients from the subjects by analyzing the collected indicators.In terms of software design,the interface of the software was designed based on Qt,the eye-movement test project was designed based on the portable eye tracker Tobii4 C,and in addition,we also designed other cognitive psychology test projects.In the aspect of algorithm design,we designed the binary classification model and the outlier detection model based on the Tensor Flow,using the data collected by software from normal subjects and patients with early AD.The binary classification model was used to distinguish between normal samples and AD samples,while the outlier detection model only learned the AD samples to distinguish the AD samples from normal samples.Results: In terms of software design,the test software can run normally and the test indicators can be obtained normally.In this paper,we collected 50 normal persons and 8patients with early AD.By statistical analysis,there were statistical differences among two groups in multiple indicators of the test items.In terms of the algorithm,the accuracy of binary classification model in distinguishing normal group from the AD samples was 97%,and the accuracy of outlier detection model to detect the AD samples was 95%.Conclusion: The Alzheimer’s disease evaluation instrument designed in this paper integrates the test software and evaluation algorithm.The subjects can obtain the screening results upon completing the test project,which provides a feasible scheme for the popularization of AD screening in general families,and has an important significance for the early detection and treatment of AD.