A Speech Prosody-Based Approach to Early Detection of Cognitive Impairment in Elderly Subjects

Abstract

This chapter presents a novel approach for early detection of cognitive impairment in the elderly. The approach incorporates the use of speech sound analysis and multivariate statistical techniques. The focus is on the prosodic features of speech. One hundred and fifteen Japanese subjects (32 males and 83 females between the ages of 38 and 99 years) participated in this study. The authors collected speech sounds from segments of dialogue during an HDS-R examination. The segments correspond to speech sounds from answers to questions about time orientation and number counting. One hundred and thirty prosodic features were extracted from each of the speech sounds. These prosodic features consisted of spectral and pitch features (53), formant features (56), intensity features (19), and speech rate and response time (2). These features were refined by principal component analysis and/or feature selection. In addition, the authors calculated speech prosody-based cognitive impairment rating (SPCIR) by multiple linear regression analysis. The results indicate that a moderately significant correlation exists between the HDS-R score and the synthesis of several selected prosodic features. Consequently, the adjusted coefficient of determination (= 0.50) suggests that prosody-based speech sound analysis could potentially be used to detect cognitive impairment in elderly subjects.