Fast and automatic assessment of fall risk by coupling machine learning algorithms with a depth camera to monitor simple balance tasks.
Amandine Dubois, Audrey Mouthon, Ranjith Steve Sivagnanaselvam and 1 others
PMID 31186002WHAT IT FOUND
Standing with eyes closed on a foam pad or in a narrow stance let a depth camera and algorithm separate older adults into higher and lower fall-risk groups.
It did not test actual falls, so use it as a screening aid, not proof.
Key findings
01Balance tasks 4 and 8, standing with eyes closed either in a narrow stance or on a foam pad, were the tasks where camera-based clusters separated older adults by Timed Up and Go performance.
02The maximum speed of the centroid was the single balance measure that gave the most consistent grouping for tasks 4 and 8.
03The clusters matched age, physical activity and Timed Up and Go scores, but the study did not observe actual falls.
STILL TO COME
How it was doneWhat they foundWhat it means for PTs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study did not follow participants to see who fell, so it cannot show that the camera method predicts future falls. The older group had low to intermediate fall risk: none had fallen and all had Timed Up and Go times under 13.5 seconds, so the method was not tested in people already at high risk. Only tasks 4 and 8 gave useful separation; tasks 1, 2, 3 and 7 did not separate the groups, and task 6 was difficult for many participants.
The easy way to misread this
Do not conclude that this depth camera system prevents or predicts falls. The study did not follow participants for falls; it only compared camera-based clusters with age, physical activity and Timed Up and Go scores in 40 older adults who had not fallen.