Detecting Early Signs of Dementia through Speech Analysis (ARD/293)

Detecting Early Signs of Dementia through Speech Analysis (ARD/293)

Detecting Early Signs of Dementia through Speech Analysis (ARD/293)
ARD/293
Seed
17 / 04 / 2023 - 16 / 04 / 2024
2,799.767

Mr Laifa FANG

Hong Kong Young Women’s Christian Association


Hong Kong is undergoing an ageing transformation. More than 100,000 people are now suffering from dementia. The number of dementia cases will increase and project to 300,000 by 2039. However, the assessment of dementia requires clinical examination supplemented by invasive laboratory tests. The diagnosis results have to rely on the clinical professional judgment. The whole process is time-consuming and expensive. This project aims to develop an automated and self-managed solution for early dementia detection via speech analysis. A Cantonese corpus will be built for dementia-related research and study. Based on the corpus, the extracted acoustic and linguistic features will be used to train an Artificial Intelligence (AI) model. The trained AI model will be applied to evaluate the category of dementia (including Mild Cognitive Impairment) by analyzing the elderly speech. The solution can be carried out periodically via mobile devices such as smartphones or tablets. It will provide a flexible way of usage without the need for frequent healthcare professional visits. The solution can be deployed on a scale for the elderly at home.