May 2, 2016
SKEL researcher, Dr Vassiliki Rentoumi will give a talk on “Machine learning classification of autopsy proven vascular cognitive impairment and Alzheimer’s disease using connected speech sampling” on Wednesday 11/5.
Abstract
This talk aims to present recent research advances in the early detection and classification of various types of dementia patients employing computational linguistics methods. Moreover, although Vascular Dementia (VaD) and Alzheimer’s Disease (AD) are associated with changes in spoken language, these have not been extensively examined or compared. In the presented study: (1) the presence of differences in language produced by AD and VaD is confirmed using quantitative methods of evaluation, and (2) the most informative sources of variation between the groups is ascertained. A computational approach was adopted for the comparison of syntactic complexity and lexical variation in digitized transcripts of speech produced by AD and VaD patients and by age-matched cognitively normal controls (NC). We used machine learning text classification approaches to assign the samples to one of the three groups – AD, VaD and NC. The classification results illustrate the specific language idiosyncrasies and the word content specific differences of the three groups. Experimental results indicate that VaDs are found to exhibit a marked reduction in lexical variation and complexity compared to their AD counterparts.