A mobile application using automatic speech analysis for classifying Alzheimer’s disease and mild cognitive impairment
Abstract
Speech and language disturbance have been observed from the early stage of Alzheimer’s disease (AD) including mild cognitive impairment (MCI), and speech analysis has been expected to help in the early detection of AD and MCI as a screening tool. However, whether and how automatic speech analysis including a speech recognition within a self-administered tool can be used for detecting AD and MCI remains largely unexplored. In this study, we developed an automatic analysis of speech data collected during cognitive tasks with a mobile application for classifying AD, MCI and cognitively normal (CN) using speech features characterizing acoustic, prosodic, and linguistic aspects, and tested it on 114 older participants. We first evaluated how accurately speech features can be automatically extracted from transcriptions generated by an automatic speech recognition and found that the features were highly correlated (r = 0.92) with those extracted from manual transcriptions. The machine-learning speech classifier by using these speech features achieved 78.6% accuracy for classifying AD, MCI, and CN through nested-cross validation (AD vs CN = 91.2% accuracy, MCI vs CN = 87.6% accuracy). Our results suggest the utility and validity of a mobile application using automatic speech analysis as a self-administered screening tool for early detection of AD and MCI.