Handwriting and drawing to predict cognitive impairnent?

In Beating the Dementia Monster, I wrote about my deteriorating handwriting.  My signature deteriorated dramatically, and sometimes I could hardly sign my name at all.  I often ripped up checks when my signature was simply unrecognizable.  Like many of the symptoms of my disease, this behavior has waxed and waned over the past seven years.  Sometimes I take the time to just practice signing my name to try getting my handwriting back.

I've written about similar issues with my voice.  In 2019, I participated in research to develop biomarkers for various neurodegenerative diseases based on human voice.  Last I heard, the research was ongoing, but it looked probable that a computer might evaluate someone's voice for clues and identify incipient disease.

So can the same be done with handwriting?  Just based on my personal experience where I can perceive a correlation between a deterioration in my ability to write and the progress of my disease, I would very definitely say yes.  There must be.

Researchers in Japan and the United States have collaborated on research to determine if they can estimate someone's score on the Montreal Cognitive Assessment (MoCA) based on a computer analysis of how they draw.  We discussed the MoCA in Beating the Dementia Monster, noting that it includes some drawing challenges, such as drawing a clock.  A perfect score is 30 points.  A score of less than 26 indicates mild cognitive impairment or worse.  People with Alzheimer's dementia may score an average of 16.  (The more commonly used mini-mental status (or state) exam includes similar challenges and is scored similarly.)

So could the researchers estimate an MoCA score?  Yes, with a respectable level of accuracy.  Their research was published in the journal JMIR Formative Research.  It used cohorts of both impaired and cognitively normal subjects in both the United States and Japan, showing similar results across varying cultures. 

So what did they measure?  Drawing speed, drawing speed variability, pauses between drawings, pen pressure, and both vertical and horizontal pen inclinations.  And what did they find?  With low MoCA scores, there tended to be higher variability in drawing speed, a higher pause duration/drawing duration ratio, and lower variability in the pen’s horizontal inclination.  From these results, they could estimate MoCA scores, apparently with greater accuracy than with the speech analysis techniques.

I know that I'd do horribly with this test.  I wonder how well it would predict my MoCA score knowing how much my memory and cognition have improved since 2015 while my handwriting hasn't.