Science & tech · 2021
Can processing pen movements during handwriting and drawing improve how a computer classifies depression, anxiety and stress?
Yes—in a study using records from 129 people, a computer model classified emotional-state categories more accurately than a baseline method. The categories were assigned using the Depression, Anxiety and Stress Scale. The authors reported an average improvement of 15%, but did not give the final accuracy or clarify how the improvement was calculated. This was a database classification study, not a test of clinical diagnosis.
Details
The researchers used tablet and pen recordings collected while people completed three handwriting tasks and four drawing tasks. The database categorized their states as depressed, anxious, stressed or typical according to the Depression, Anxiety and Stress Scale (DASS).
The computer model combined measurements of pen movement, pressure and timing with features describing patterns in the recorded signals. The researchers selected useful features and expanded the training data by adding small random changes to some existing measurements. They evaluated the model by leaving out each person in turn for testing and training on the remaining people.
Compared with a baseline method described as state of the art, the authors reported an average classification-accuracy improvement of 15%, with improvements ranging from 4% to 34% across the emotional-state categories. The abstract does not report final accuracy values or clearly define how those improvements were calculated. The findings support improved classification within this small database, not a demonstrated ability to diagnose mental health conditions in new populations.
What do you think?
Good conversations begin with curiosity.
You can read comments without an account. Sign in to write.
Loading comments…