شاخص معنایی خودکار مربوط به عملکرد شناختی و میزان کاهش شناختی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|76041||2012||11 صفحه PDF||سفارش دهید||8886 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Neuropsychologia, Volume 50, Issue 9, July 2012, Pages 2165–2175
The objective of our study is to introduce a fully automated, computational linguistic technique to quantify semantic relations between words generated on a standard semantic verbal fluency test and to determine its cognitive and clinical correlates. Cognitive differences between patients with Alzheimer’s disease and mild cognitive impairment are evident in their performance on the semantic verbal fluency test. In addition to the semantic verbal fluency test score, several other performance characteristics sensitive to disease status and predictive of future cognitive decline have been defined in terms of words generated from semantically related categories (clustering) and shifting between categories (switching). However, the traditional assessment of clustering and switching has been performed manually in a qualitative fashion resulting in subjective scoring with limited reproducibility and scalability. Our approach uses word definitions and hierarchical relations between the words in WordNet®, a large electronic lexical database, to quantify the degree of semantic similarity and relatedness between words. We investigated the novel semantic fluency indices of mean cumulative similarity and relatedness between all pairs of words regardless of their order, and mean sequential similarity and relatedness between pairs of adjacent words in a sample of patients with clinically diagnosed probable (n=55) or possible (n=27) Alzheimer’s disease or mild cognitive impairment (n=31). The semantic fluency indices differed significantly between the diagnostic groups, and were strongly associated with neuropsychological tests of executive function, as well as the rate of global cognitive decline. Our results suggest that word meanings and relations between words shared across individuals and computationally modeled via WordNet and large text corpora provide the necessary context to account for the variability in language-based behavior and relate it to cognitive dysfunction observed in mild cognitive impairment and Alzheimer’s disease.