'Computer avatar may help detect dementia'

Japanese scientists say they have developed a novel machine-learning technique that can detect dementia with over 90 per cent accuracy from conversations between humans and avatars on a computer.

This technique developed by a team from Osaka University and Nara Institute of Science and Technology involves learning characteristics of sounds of elderly people who answer easy questions.

Dementia could be distinguished by combining features of the disorder, such as delay in response to questions from avatars depending on the content of questions, intonation, articulation rate of the voice, and the percentage of nouns and verbs in utterance.

An avatar is the graphical representation of the user or a character in an internet forum.

 

"If this technology is further developed, it will become possible to know whether or not an elderly individual is in the early stages of dementia through conversation with computer avatars at home on a daily basis," said researcher Takashi Kudo.

"It will encourage them to seek medical help, leading to early diagnosis," Takashi said.

The researchers proposed machine learning algorithms for detecting signs of dementia in its early stages, using interactive computer avatars.

In the research published in IEEE Journal of Translational Engineering in Health and Medicine, the team created a model for machine learning based on features of speech, language, and faces from recorded dialogues with elderly participants.

Through machine learning, a computer was able to distinguish individuals with dementia from healthy controls at a rate of 90 per cent in six questions, researchers said.

The team prepared fixed questions based on neuropsychological tests and random questions not based on specific tests.

The process involved recording interactive data of spoken dialogues with avatars from 12 individuals diagnosed with dementia and 12 healthy controls.

The researchers extracted speech, language, and image features from the recorded data, creating a model for detecting dementia and enabling a computer to learn for itself to detect dementia.

As a result, the computer was able to distinguish between healthy controls and individuals with dementia with an accuracy of 92 per cent, researchers said.