Hidden Markov Model based stride segmentation on unsupervised free-living gait data in Parkinson's disease patients.
HMM
IMU
Machine learning
Mobile gait analysis
Stride borders
Wearable sensors
Journal
Journal of neuroengineering and rehabilitation
ISSN: 1743-0003
Titre abrégé: J Neuroeng Rehabil
Pays: England
ID NLM: 101232233
Informations de publication
Date de publication:
03 06 2021
03 06 2021
Historique:
received:
24
02
2021
accepted:
20
05
2021
entrez:
4
6
2021
pubmed:
5
6
2021
medline:
26
11
2021
Statut:
epublish
Résumé
To objectively assess a patient's gait, a robust identification of stride borders is one of the first steps in inertial sensor-based mobile gait analysis pipelines. While many different methods for stride segmentation have been presented in the literature, an out-of-lab evaluation of respective algorithms on free-living gait is still missing. To address this issue, we present a comprehensive free-living evaluation dataset, including 146.574 semi-automatic labeled strides of 28 Parkinson's Disease patients. This dataset was used to evaluate the segmentation performance of a new Hidden Markov Model (HMM) based stride segmentation approach compared to an available dynamic time warping (DTW) based method. The proposed HMM achieved a mean F1-score of 92.1% and outperformed the DTW approach significantly. Further analysis revealed a dependency of segmentation performance to the number of strides within respective walking bouts. Shorter bouts ([Formula: see text] strides) resulted in worse performance, which could be related to more heterogeneous gait and an increased diversity of different stride types in short free-living walking bouts. In contrast, the HMM reached F1-scores of more than 96.2% for longer bouts ([Formula: see text] strides). Furthermore, we showed that an HMM, which was trained on at-lab data only, could be transferred to a free-living context with a negligible decrease in performance. The generalizability of the proposed HMM is a promising feature, as fully labeled free-living training data might not be available for many applications. To the best of our knowledge, this is the first evaluation of stride segmentation performance on a large scale free-living dataset. Our proposed HMM-based approach was able to address the increased complexity of free-living gait data, and thus will help to enable a robust assessment of stride parameters in future free-living gait analysis applications.
Sections du résumé
BACKGROUND
To objectively assess a patient's gait, a robust identification of stride borders is one of the first steps in inertial sensor-based mobile gait analysis pipelines. While many different methods for stride segmentation have been presented in the literature, an out-of-lab evaluation of respective algorithms on free-living gait is still missing.
METHOD
To address this issue, we present a comprehensive free-living evaluation dataset, including 146.574 semi-automatic labeled strides of 28 Parkinson's Disease patients. This dataset was used to evaluate the segmentation performance of a new Hidden Markov Model (HMM) based stride segmentation approach compared to an available dynamic time warping (DTW) based method.
RESULTS
The proposed HMM achieved a mean F1-score of 92.1% and outperformed the DTW approach significantly. Further analysis revealed a dependency of segmentation performance to the number of strides within respective walking bouts. Shorter bouts ([Formula: see text] strides) resulted in worse performance, which could be related to more heterogeneous gait and an increased diversity of different stride types in short free-living walking bouts. In contrast, the HMM reached F1-scores of more than 96.2% for longer bouts ([Formula: see text] strides). Furthermore, we showed that an HMM, which was trained on at-lab data only, could be transferred to a free-living context with a negligible decrease in performance.
CONCLUSION
The generalizability of the proposed HMM is a promising feature, as fully labeled free-living training data might not be available for many applications. To the best of our knowledge, this is the first evaluation of stride segmentation performance on a large scale free-living dataset. Our proposed HMM-based approach was able to address the increased complexity of free-living gait data, and thus will help to enable a robust assessment of stride parameters in future free-living gait analysis applications.
Identifiants
pubmed: 34082762
doi: 10.1186/s12984-021-00883-7
pii: 10.1186/s12984-021-00883-7
pmc: PMC8173987
doi:
Types de publication
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
93Références
Gait Posture. 2012 Jun;36(2):316-8
pubmed: 22465705
PLoS One. 2021 Feb 4;16(2):e0246528
pubmed: 33539481
Int J Environ Res Public Health. 2015 Jun 29;12(7):7274-99
pubmed: 26132480
Lancet Neurol. 2019 Jul;18(7):697-708
pubmed: 30975519
J Biomech. 2002 May;35(5):689-99
pubmed: 11955509
Physiol Meas. 2017 Jan;38(1):N1-N15
pubmed: 27941238
Annu Int Conf IEEE Eng Med Biol Soc. 2011;2011:4369-73
pubmed: 22255307
Annu Int Conf IEEE Eng Med Biol Soc. 2015;2015:5179-82
pubmed: 26737458
Mov Disord. 2016 Sep;31(9):1293-313
pubmed: 27452964
J Gerontol A Biol Sci Med Sci. 2019 Mar 14;74(4):500-506
pubmed: 29300849
IEEE J Biomed Health Inform. 2020 Jul;24(7):1879-1886
pubmed: 32386168
Gait Posture. 2012 Sep;36(4):657-61
pubmed: 22796244
Sensors (Basel). 2015 Mar 17;15(3):6419-40
pubmed: 25789489
IEEE Trans Biomed Eng. 2015 Apr;62(4):1089-97
pubmed: 25389237
Annu Int Conf IEEE Eng Med Biol Soc. 2019 Jul;2019:1834-1837
pubmed: 31946254
PLoS One. 2017 Oct 11;12(10):e0183989
pubmed: 29020012
Diseases. 2019 Feb 05;7(1):
pubmed: 30764502
Sensors (Basel). 2019 Apr 16;19(8):
pubmed: 30995789
Annu Int Conf IEEE Eng Med Biol Soc. 2018 Jul;2018:2284-2287
pubmed: 30440862
IEEE J Biomed Health Inform. 2020 Jul;24(7):1869-1878
pubmed: 32086225
Front Neurol. 2020 Apr 21;11:261
pubmed: 32373047
Sensors (Basel). 2019 Jul 13;19(14):
pubmed: 31337067
PLoS One. 2018 May 1;13(5):e0196463
pubmed: 29715279
IEEE J Biomed Health Inform. 2020 May;24(5):1490-1499
pubmed: 31449035
J Biomech. 2008;41(1):216-20
pubmed: 17897652
IEEE Trans Biomed Eng. 2020 Aug;67(8):2132-2144
pubmed: 31765301
J Neuroeng Rehabil. 2019 Jun 26;16(1):77
pubmed: 31242915
J Neuroeng Rehabil. 2016 May 12;13(1):46
pubmed: 27175731
Sensors (Basel). 2018 Jan 06;18(1):
pubmed: 29316636
Gait Posture. 2018 Jan;59:248-252
pubmed: 29100144