With the continuous intensification of population aging, degenerative cognitive impairment diseases represented by Alzheimer’s disease have imposed a heavy burden on society and families. Developing simple and rapid methods for detecting Alzheimer’s disease or its early symptoms has emerged as a research hotspot in the academic community. Screening potential Alzheimer’s patients based on language features can significantly reduce diagnostic costs and predict disease risks. Along with the rapid development of technologies such as natural language processing and machine learning, a faster, more accurate, and large-scale automated method based on artificial intelligence can be employed to analyze language sample data. Existing research has indicated that, in contrast to single-word sentences, studying the coherent language of patients enables more effective analysis of the language features of patients with cognitive impairment, thereby facilitating more accurate identification of language markers related to cognitive decline. This study utilizes the Chinese natural language processing tool LTP to analyze the Chinese coherent spoken language samples of 51 elderly people with dementia and 48 healthy young people. A dataset of lexical and syntactic language features based on coherent discourse is constructed, and machine learning classification algorithms are trained and applied to classification tasks. The experimental results demonstrate that machine learning classification algorithms can effectively distinguish the cognitive abilities of the Chinese-speaking population based on lexical and syntactic language features in coherent discourse.
Chen et al. (Fri,) studied this question.