Improving Accelerometry-Based Measurement of Functional Use of the Upper Extremity After Stroke: Machine Learning Versus Counts Threshold Method.
accelerometry
machine learning
neurorehabilitation
stroke
upper extremity
Journal
Neurorehabilitation and neural repair
ISSN: 1552-6844
Titre abrégé: Neurorehabil Neural Repair
Pays: United States
ID NLM: 100892086
Informations de publication
Date de publication:
12 2020
12 2020
Historique:
pubmed:
6
11
2020
medline:
6
10
2021
entrez:
5
11
2020
Statut:
ppublish
Résumé
Wrist-worn accelerometry provides objective monitoring of upper-extremity functional use, such as reaching tasks, but also detects nonfunctional movements, leading to ambiguity in monitoring results. Compare machine learning algorithms with standard methods (counts ratio) to improve accuracy in detecting functional activity. Healthy controls and individuals with stroke performed unstructured tasks in a simulated community environment (Test duration = 26 ± 8 minutes) while accelerometry and video were synchronously recorded. Human annotators scored each frame of the video as being functional or nonfunctional activity, providing ground truth. Several machine learning algorithms were developed to separate functional from nonfunctional activity in the accelerometer data. We also calculated the counts ratio, which uses a thresholding scheme to calculate the duration of activity in the paretic limb normalized by the less-affected limb. The counts ratio was not significantly correlated with ground truth and had large errors ( In our sample, the counts ratio did not accurately reflect functional activity. Machine learning algorithms were more accurate, and future work should focus on the development of a clinical tool.
Sections du résumé
BACKGROUND
Wrist-worn accelerometry provides objective monitoring of upper-extremity functional use, such as reaching tasks, but also detects nonfunctional movements, leading to ambiguity in monitoring results.
OBJECTIVE
Compare machine learning algorithms with standard methods (counts ratio) to improve accuracy in detecting functional activity.
METHODS
Healthy controls and individuals with stroke performed unstructured tasks in a simulated community environment (Test duration = 26 ± 8 minutes) while accelerometry and video were synchronously recorded. Human annotators scored each frame of the video as being functional or nonfunctional activity, providing ground truth. Several machine learning algorithms were developed to separate functional from nonfunctional activity in the accelerometer data. We also calculated the counts ratio, which uses a thresholding scheme to calculate the duration of activity in the paretic limb normalized by the less-affected limb.
RESULTS
The counts ratio was not significantly correlated with ground truth and had large errors (
CONCLUSIONS
In our sample, the counts ratio did not accurately reflect functional activity. Machine learning algorithms were more accurate, and future work should focus on the development of a clinical tool.
Identifiants
pubmed: 33150830
doi: 10.1177/1545968320962483
pmc: PMC7704838
mid: NIHMS1629315
doi:
Types de publication
Comparative Study
Journal Article
Research Support, Non-U.S. Gov't
Research Support, U.S. Gov't, Non-P.H.S.
Langues
eng
Sous-ensembles de citation
IM
Pagination
1078-1087Subventions
Organisme : ACL HHS
ID : 90REGE0004
Pays : United States
Références
PLoS One. 2019 Mar 20;14(3):e0212484
pubmed: 30893308
Neurorehabil Neural Repair. 2012 Jan;26(1):76-84
pubmed: 21693771
Arch Phys Med Rehabil. 2013 Jan;94(1):86-94
pubmed: 22922823
Int Rehabil Med. 1986;8(2):69-73
pubmed: 3804600
Am J Occup Ther. 2016 May-Jun;70(3):7003290040p1-7003290040p11
pubmed: 27089298
Arch Phys Med Rehabil. 2016 Feb;97(2):224-31
pubmed: 26435302
Neurology. 2006 Oct 10;67(7):1189-94
pubmed: 17030751
Arch Phys Med Rehabil. 1979 Nov;60(11):487-91
pubmed: 508073
Med Sci Sports Exerc. 2017 Nov;49(11):2351-2360
pubmed: 28604558
Neuropsychologia. 1971 Mar;9(1):97-113
pubmed: 5146491
Cerebrovasc Dis. 2008;26(5):533-40
pubmed: 18836264
J Stroke Cerebrovasc Dis. 2017 Dec;26(12):2880-2887
pubmed: 28781056
Stroke. 2002 Jul;33(7):1840-4
pubmed: 12105363
IEEE J Biomed Health Inform. 2014 Nov;18(6):1804-12
pubmed: 25014974
J Phys Ther Sci. 2014 Jul;26(7):1009-11
pubmed: 25140084
Neurorehabil Neural Repair. 2015 Nov-Dec;29(10):969-78
pubmed: 25896988
Arch Phys Med Rehabil. 2005 Jul;86(7):1498-501
pubmed: 16003690
Arch Phys Med Rehabil. 2018 Sep;99(9):1913-1916
pubmed: 29408483
J Neurol Phys Ther. 2007 Jun;31(2):56-63
pubmed: 17558358
Neurorehabil Neural Repair. 2017 Mar;31(3):290-300
pubmed: 27909071
Arch Phys Med Rehabil. 2012 Apr;93(4):660-8
pubmed: 22336104
Neurology. 2009 Jul 21;73(3):195-201
pubmed: 19458319
Curr Neurol Neurosci Rep. 2018 Oct 6;18(12):87
pubmed: 30293160
J Rehabil Med. 2001 Mar;33(3):110-3
pubmed: 11482350
J Rehabil Med. 2011 Mar;43(4):299-304
pubmed: 21347506
Clin Rehabil. 2012 Feb;26(2):111-20
pubmed: 21840917
Arch Phys Med Rehabil. 2011 Sep;92(9):1437-42
pubmed: 21878214
Stroke. 2005 Nov;36(11):2493-6
pubmed: 16224078
Rehabil Psychol. 2009 Nov;54(4):398-403
pubmed: 19929121
Stroke. 2000 Mar;31(3):662-7
pubmed: 10700501
Lancet Neurol. 2009 Aug;8(8):741-54
pubmed: 19608100
Arch Phys Med Rehabil. 2010 Feb;91(2):288-97
pubmed: 20159136
J Neurol Neurosurg Psychiatry. 1983 Jun;46(6):521-4
pubmed: 6875585
Circulation. 2018 Mar 20;137(12):e67-e492
pubmed: 29386200
J Neuroeng Rehabil. 2014 Oct 09;11:144
pubmed: 25297823
Med Sci Sports Exerc. 2015 Oct;47(10):2129-39
pubmed: 25785929
Neurology. 1988 Oct;38(10):1565-8
pubmed: 3419600
Lancet. 2015 Jul 4;386(9988):46-55
pubmed: 25892679
Arch Phys Med Rehabil. 2006 Oct;87(10):1340-5
pubmed: 17023243
Clin Rehabil. 2018 Jan;32(1):75-83
pubmed: 28580791
Stroke. 2001 Jun;32(6):1279-84
pubmed: 11387487
Stroke. 1999 Oct;30(10):2131-40
pubmed: 10512918
PLoS One. 2014 Feb 04;9(2):e87987
pubmed: 24505342
Physiol Meas. 2014 Dec;35(12):2359-67
pubmed: 25389908
Assist Technol. 2019;31(4):193-198
pubmed: 29215963
Arch Phys Med Rehabil. 2006 Dec;87(12):1605-10
pubmed: 17141640
Arch Phys Med Rehabil. 2015 May;96(5):854-61
pubmed: 25497517