Using Machine Learning for Recognition of Alzheimer’s Disease Based on Transcription Information

Abstract

The purpose of this article is to perform analytical and prognostic studies on the recognition of Alzheimer’s disease based on decoded text speech data using machine learning algorithms. The data used in this article is taken from the ADReSS 2020 Challenge program, which contains speech data from patients with Alzheimer’s disease and healthy people. The problem under study is a binary classification problem. First, the full texts of the interviewees were extracted from the transcribed texts of the speech data. This was followed by training the model based on vectorized text features using a random forest classifier, in which the authors used the GridSearchCV method to optimize hyperparameters. The classification accuracy of the model reached 85.2 %.