Automatic Prediction of Linguistic Decline in Writings of Subjects with Degenerative Dementia

Proceedings of NAACL-HLT 2016, pages 1198–1207, San Diego, California, June 12-17, 2016. c⃝2016 Association for Computational Linguistics Towards Automatic Detection of Abnormal Cognitive Decline and Dementia Through Linguistic Analysis of Writing Samples Weissenbacher Davy Department of Biomedical Informatics, ASU Scottsdale, AZ, USA davy.weissen- bacher@asu.edu Johnson A. Travis Department of Neurology Mayo Clinic Scottsdale, AZ, USA Wojtulewicz Laura Department of Biomedical Informatics, ASU Scottsdale, AZ, USA Dueck Amylou Department of Biostatistics Mayo Clinic Scottsdale, AZ, USA Locke Dona Department of Psychiatry and Psychology, Mayo Clinic Scottsdale, AZ, USA Caselli Richard Department of Neurology Mayo Clinic Scottsdale, AZ, USA Gonzalez Graciela Department of Biomedical Informatics, ASU Scottsdale, AZ, USA Graciela.Gon- zalez@asu.edu Abstract Given the limited success of medication in re- versing the effects of Alzheimer’s and other dementias, a lot of the neuroscience research has been focused on early detection, in order to slow the progress of the disease through different interventions. We propose a Natural Language Processing approach applied to de- scriptive writing to attempt to discriminate decline due to normal aging from decline due to pre-dementia conditions. Within the context of a longitudinal study on Alzheimer’s dis- ease, we created a unique corpus of 201 de- scriptions of a control image written by sub- jects of the study. Our classifier, computing linguistic features, was able to discriminate normal from cognitively impaired patients to an accuracy of 86.1% using lexical and se- mantic irregularities found in their writing. This is a promising result towards elucidating the existence of a general pattern in linguistic deterioration caused by dementia that might be detectable from a subject’s written descrip- tive language. 1 Introduction Alzheimer’s disease is prevalent and becoming more so as the world’s population ages (Prince et al., 2014). Since no cure is known, it is hoped that early detection and intervention might slow the on- set of symptomatic cognitive decline and dementia. Clinical methods to detect Alzheimer’s disease are typically applied well after symptoms have pro- gressed to a troubling degree, and may be costly. Families, however, often report earlier signs of the disease through their language interactions with their elders. This has led clinical researchers to study linguistic differences to detect the disease in conversational speech (Asp and de Villiers, 2010). One approach is to search for non-informative phrases or semantic incoherences, which was con- firmed to distinguish patients with Alzheimer’s disease from controls (Nicholas et al., 1985). A strong limitation for its automatic application is the need of a trained expert to annotate the incoher- ences and scoring by hand. We propose in this study to use Natural Lan- guage Processing (NLP) to evaluate samples of a patient’s descriptive writing in order to attempt to 1198 discriminate decline due to normal aging from de- cline due to pre-demented conditions. The Arizona Alzheimer’s Disease Center (ADC) is a longitudi- nal study of patients with Alzheimer’s disease and normal control subjects, who receive an annual battery of clinical and neuropsychological exams, to which we added the following brief and a simple task. …

Content truncated to 500 words