Digital Analytics on Simple Drawing Is An Effective Method for Dementia Screening
Refereed conference paper presented and published in conference proceedings

Altmetrics Information

Other information
Drawing, such as a pair of interlocking pentagons, is a useful measurement to evaluate visuospatial abilities and executive functions for people with cognitive impairment. With advancement of technology, all personal drawing behaviour can be real-time captured with digital platforms. Digital Clock Drawing Test is a typical case of technological application with real-time capturing on clock drawing. However, drawing a clock may be difficult for those who did not officially receive education. We proposed digital analytics on a simple drawing platform, and hypothesized that this platform can be used for preliminary screening for dementia.

Patients with Alzheimer’s disease (AD) and Montreal Cognitive Assessment (MoCA) score below 22 were recruited from the Geriatric Research Clinic, and subjects without AD and with MoCA score at least 22 were recruited from the Osteoporosis Research Centre. An automated scoring platform for interlocking pentagons was developed. All drawing behaviour was digitalized as spatial and temporal data, such as the time of drawing for each line, hesitation moments between the drawing lines. Some features of the drawing image were automatically identified, including overlapped and closure pentagons, asymmetric shape, and irregular line by a shaking hand. All time and image features between AD and control were compared by T-test or Chi-square test. Multivariate logistic regression model was fitted as the predictors on AD.

A total of 93 subjects were recruited, and 67 of them (72%) were AD patients. The mean age of AD patients was 79.7 (SD=5.4), and with average 11.2 MoCA scores (SD=4.9). In the 26 subjects without AD, the mean age was 82.7 (SD=3.4), and with average 24.4 MoCA scores (SD=1.7). Compared with control subjects, AD patients are less capable of drawing 10 angles in the interlocking pentagon (20.9% vs 57.7%) (OR=4.86 95% CI [1.47-16.12], p=0.010); and longer drawing time on the first pentagon (4.7 vs 3.8 seconds) (OR=2.09 95% CI [1.25-3.50], p=0.005) [Table 1].

Analytics on the digital behaviour of simple drawing is an effective method for dementia screening. Further investigation with machine learning techniques may increase data interpretability and potentially enhance the diagnostic accuracy for dementia screening.
All Author(s) ListKelvin Tsoi, Michael PF. Wong, Max WY. Lam, Adrian Wong, Timothy Kwok
Name of ConferenceAlzheimer’s Association International Conference (AAIC) 2017
Start Date of Conference16/07/2017
End Date of Conference20/07/2017
Place of ConferenceLondon
Country/Region of ConferenceGreat Britain
Proceedings TitleAlzheimer's & Dementia: The Journal of the Alzheimer's Association - 2017 Abstract Supplement
Series TitlePoster Presentations: Saturday, July 15, 2017
Number in SeriesTD-P-017
Volume Number13
Issue Number7 Supplement
Pages165 - 165
LanguagesEnglish-United Kingdom

Last updated on 2020-19-10 at 01:59