Assessing Alzheimer's Disease from Speech Using the i-vector Approach.

摘要: One of the world’s chronic neuro-degenerative diseases, Alzheimer’s Disease (AD), leads its sufferers, among other symptoms, to suffer from speech difficulties. In particular, inability recall vocabulary which makes patients’ different. Furthermore, Mild Cognitive Impairment (MCI) is usually considered as a prodromal state AD. The key abate progress both disorders their early diagnosis. However, actual ways diagnosis are costly and quite time-consuming. this study, we propose extraction features through use i-vector approach, by seek model pattern three mental conditions subjects. To best our knowledge, no previous studies have utilized assess before. These i-vectors extracted Mel-Frequency Cepstral Coefficients (MFCCs), then they given SVM classifier in order identify one following manners: AD - Alzheimer Disease, MCI Impairment, HC Healthy Control. We tested these performing 5-fold cross-validation achieved an F1-score 79.2%.