Deep learning can accurately distinguish dementia from mild cognitive impairment by looking at pictures of analog clocks as drawn by individuals thought to be affected.
So found computer engineers at Boston University when they trained and tested a deep learning algorithm on more than 3,200 clock drawings created by mentally healthy persons and 160 from peers in various states of cognitive decline.
All subjects were participants in the Framingham Heart Study.
For one arm of the study, the researchers applied their algorithm only to the pictures.
For the other, they added annotations of names, ages and educational levels.
On analysis of the picture-only approach, the researchers found their model predicted dementia status with an area under the receiver operating characteristic curve (AUC) of 81.3%.
However, when they added the annotated info to the drawing-only predictions, the model’s performance jumped to 91.9% AUC.
The clock drawing test (CDT) has been used for decades to screen for cognitive impairment.
The research augmenting interpretations with deep learning was posted July 27 in the Journal of Alzheimer’s Disease.
Senior author Ioannis Paschalidis, PhD, and co-authors comment:
Our modeling framework establishes a proof-of-principle that deep learning can be applied on images derived from the CDT to predict dementia status. When fully validated, this approach can offer a cost-effective and easily deployable mechanism for detecting cognitive impairment.”
Lead author is PhD candidate Samad Amini.
Study abstract here (full report behind paywall).
Dave P. has worked in journalism, marketing and public relations for more than 30 years, frequently concentrating on hospitals, healthcare technology and Catholic communications. He has also specialized in fundraising communications, ghostwriting for CEOs of local, national and global charities, nonprofits and foundations.