Development of a novel activity index that predicts changes in cognitive function
Principal Investigator: Dr Dukyong Yoon
Approved Research ID: 40882
Approval date: August 30th 2019
Aims: The prediction of changes in cognitive function will enable the early diagnosis and prevention of dementia. Several lines of evidence support the use of actigraphic data to evaluate cognitive function. However, studies that have aimed to predict changes in cognitive function based on actigraphic data are lacking. We aim to investigate the usefulness of existing activity indexes, such as circadian rhythm, high-intensity activity rate, and activity complexity, to evaluate their relationships with cognitive function, and to identify a novel activity index that can be used to predict changes in cognitive function. Scientific rationale: Two studies support the use of actigraphic data to predict changes in cognitive function. Varma and Watts showed that individuals with low cognitive function have low rates of high-intensity activity and low levels of physical activity complexity. Tranah et al. suggested a relationship between activity patterns and the development of dementia; the risk of dementia in older women was increased in subjects with unstable circadian rhythms. Previous studies have used activity indicators, such as daily rhythm, high-intensity activity rate, and activity complexity, to identify relationships between activity patterns and cognitive function. However, we aim to discover a novel activity index by analysing the relationships between activity characteristics and changes in cognitive function using deep learning (a machine learning method that does not require manually curated features). Project duration: about 2 years - Until April 2019: data preparation - April ~ December 2019: data pre-processing and evaluation of existing activity indexes - January ~ June 2020: discovery of a novel activity index using the deep learning model - July ~ December 2020: subgroup analysis - January ~ July 2021: summary and supplement of research results Expected impact on the public health: Our study will contribute to discover novel activity index which can be used to predict cognitive function changes. It will be able to be used for early detection and prevention of dementia. Specifically, activity index can be used to monitor high-risk patients with dementia non-invasively and continuously using mobile devices. Furthermore, intervention strategy on activity of patient to prevent or delay progress of dementia can be established.