Identifying neurocognitive disorder using vector representation of free conversation

Abstract

In recent years, studies on the use of natural language processing (NLP) approaches to identify dementia have been reported. Most of these studies used picture description tasks or other similar tasks to encourage spontaneous speech, but the use of free conversation without requiring a task might be easier to perform in a clinical setting. Moreover, free conversation is unlikely to induce a learning effect. Therefore, the purpose of this study was to develop a machine learning model to discriminate subjects with and without dementia by extracting features from unstructured free conversation data using NLP. We recruited patients who visited a specialized outpatient clinic for dementia and healthy volunteers. Participants’ conversation was transcribed and the text data was decomposed from natural sentences into morphemes by performing a morphological analysis using NLP, and then converted into real-valued vectors that were used as features for machine learning. A total of 432 datasets were used, and the resulting machine learning model classified the data for dementia and non-dementia subjects with an accuracy of 0.900, sensitivity of 0.881, and a specificity of 0.916. Using sentence vector information, it was possible to develop a machine-learning algorithm capable of discriminating dementia from non-dementia subjects with a high accuracy based on free conversation.

Actual Clinical Practice Assessment: A Rapid and Easy-to-Use Tool for Evaluating Cognitive Decline Equivalent to Dementia

“…One study found that using a convolutional neural network model and voice recordings from neuropsychological tests could detect dementia in normal individuals with up to 74.0% accuracy [ 11 ]. Another study reported that 3- to 5-minute speech recordings taken from a free conversation could be used to classify dementia versus non-dementia with an accuracy of 90.0% using a machine learning tool [ 12 ]. Thus, ONSEI has the potential to achieve detection accuracy as good as or even better than that obtained in previous studies, with a short test time of 20 seconds.…”

Applications of artificial intelligence in dementia

“…In an initial study conducted at the Department of Psychiatry, Keio University, the system exhibited a 90.0% accuracy rate in discriminating between healthy individuals, those with MCI, and those with dementia. 20 Currently, the system focuses on distinguishing healthy individuals/those with MCI from those with dementia from a medical device perspective. This suggests that the potential of the system goes beyond dementia detection and may extend to the detection of MCI as well.…”

Novel Screening Tool Using Voice Features Derived from Simple, Language-independent Phrases to Detect Mild Cognitive Impairment and Dementia

“…Another common method is a machine-learning model with linguistic features that primarily uses natural language processing (NLP) (10,11,22,23). Although these methods offer high performance in dementia detection, their linguistic features are highly language-dependent.…”