Tell me more about how your bots will cure loneliness, Zuck.

My colleague @leiashum.bsky.social will be presenting this work at the #AAIC2025 session at 9am Wednesday "Artificial Intelligence-Focused Advancements in ADRD Diagnosis and Care"

Hope to see you there!

This study supports that there is valuable information captured by the location tracking system, and that this information can be leveraged to provide clinical insights about neuropsychiatric symptoms.

We also found that in those instances when the algorithm was confident but "wrong," most of those shifts had clinical labels on the edge between agitation or non-agitation; or when they were blindly relabelled from the clinical documentation, it turned out the algorithm was actually "right".

We found that markers such as speed, length and tortuosity of path, and number of angles in the path, among others, helped to distinguish agitation from non-agitated movements.

In our recently published study, we identified digital biomarkers of motor agitation using location tracking data from a real-time location safety system installed in a dementia care unit, and used them to predict the presence of motor agitation during a nursing shift.