부산대학교 도서관

This paper proposes a dementia screening system based on spontaneous speech analysis to assist general practitioners. Our research aims to detect dementia, especially frontotemporal lobar degeneration (FTLD). We propose a new method of dementia detection by casual talk during a clinical interview that is simpler and quicker than previous studies. The method uses speech features from an answer to the question to automatically detect dementia and classify dementia types. We recruited 136 Japanese subjects (50 males and 86 females between the ages of 45 and 84) in this study. Two types of dementia patients, FTLD and Alzheimer’s disease (AD), were recruited. We used features from an answer to a clinical interview to classify and confirm that our method can detect 77% of people with dementia.