About Me
I am currently a postdoctoral fellow at the NICHE Research Group at Dalhousie University. Previously, I was a postdoctoral fellow (2011-2013) at the BFO team, ICube Laboratory at University of Strasbourg. My PhD dissertation, defended in 2011, was on an activity recognition approach based the possibility theory and description logic. The PhD was carried out at the DOMUS Laboratory at Université de Sherbrooke.
Research Interests
My main research interests include:
- Ambient intelligence and Context awareness
- Knowledge representation and Semantic Web
- Health informatics and Cognitive assistance
Currently, I am doing research on knowledge-driven activity recognition and health informatics.
Publications
2022
Kathryn Young-Shand; Patrice C. Roy; Michael Dunbar; Syed Sibte Raza Abidi; J. Wilson
Assessing Knee Osteoarthritis Severity and Biomechanical Changes After Total Knee Arthroplasty Using Self-Organizing Maps Proceedings Article
In: 20th International Conference on Artificial Intelligence in Medicine (AIME 2022), June 14-17, 2022, Halifax, Springer, 2022.
BibTeX | Tags: Self-Organizing Maps, Total knee arthroplasty
@inproceedings{nokey,
title = {Assessing Knee Osteoarthritis Severity and Biomechanical Changes After Total Knee Arthroplasty Using Self-Organizing Maps},
author = {Kathryn Young-Shand and Patrice C. Roy and Michael Dunbar and Syed Sibte Raza Abidi and J. Wilson},
year = {2022},
date = {2022-06-14},
urldate = {2022-06-14},
booktitle = {20th International Conference on Artificial Intelligence in Medicine (AIME 2022), June 14-17, 2022, Halifax},
publisher = {Springer},
keywords = {Self-Organizing Maps, Total knee arthroplasty},
pubstate = {published},
tppubtype = {inproceedings}
}
Kathryn L. Young-Shand; Patrice C. Roy; Michael J. Dunbar; Syed S. R. Abidi; Janie L. Astephen Wilson
Gait biomechanics phenotypes among total knee arthroplasty candidates by machine learning cluster analysis Journal Article
In: Journal of Orthopaedic Research, 2022.
Abstract | Links | BibTeX | Tags: gait analysis, knee biomechanics, Machine Learning, phenotypes, Total knee arthroplasty
@article{https://doi.org/10.1002/jor.25363,
title = {Gait biomechanics phenotypes among total knee arthroplasty candidates by machine learning cluster analysis},
author = {Kathryn L. Young-Shand and Patrice C. Roy and Michael J. Dunbar and Syed S. R. Abidi and Janie L. Astephen Wilson},
url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/jor.25363},
doi = {https://doi.org/10.1002/jor.25363},
year = {2022},
date = {2022-05-10},
urldate = {2022-05-10},
journal = {Journal of Orthopaedic Research},
abstract = {Abstract Knee osteoarthritis patient phenotyping is relevant to developing targeted treatments and assessing the treatment efficacy of total knee arthroplasty (TKA). This study aimed to identify clusters among TKA candidates based on demographic and knee mechanic features during gait, and compare gait changes between clusters postoperatively. TKA patients underwent 3D gait analysis 1-week pre (n = 134) and 1-year post-TKA (n = 105). Principal component analysis was applied to frontal and sagittal knee angle and moment waveforms, extracting major patterns of variability. Age, sex, body mass index, gait speed, and frontal and sagittal pre-TKA angle and moment PC scores previously identified as relevant to TKA outcomes were standardized (mean = 0},
keywords = {gait analysis, knee biomechanics, Machine Learning, phenotypes, Total knee arthroplasty},
pubstate = {published},
tppubtype = {article}
}
Abstract Knee osteoarthritis patient phenotyping is relevant to developing targeted treatments and assessing the treatment efficacy of total knee arthroplasty (TKA). This study aimed to identify clusters among TKA candidates based on demographic and knee mechanic features during gait, and compare gait changes between clusters postoperatively. TKA patients underwent 3D gait analysis 1-week pre (n = 134) and 1-year post-TKA (n = 105). Principal component analysis was applied to frontal and sagittal knee angle and moment waveforms, extracting major patterns of variability. Age, sex, body mass index, gait speed, and frontal and sagittal pre-TKA angle and moment PC scores previously identified as relevant to TKA outcomes were standardized (mean = 0
2019
Kathryn Young-Shand; Patrice C. Roy; Syed Sibte Raza Abidi; Michael Dunbar; Janie Astephen Wilson
Clinical and Biomechanical Cluster Classification Before TKA Impacts Functional Outcome Proceedings Article
In: 2nd International Combined Meeting of Orthopaedic Research Societies, June 19-22, 2019, Montreal, Canada, 2019.
Abstract | BibTeX | Tags: Clustering, Total knee arthroplasty
@inproceedings{Young-Shand2019b,
title = {Clinical and Biomechanical Cluster Classification Before TKA Impacts Functional Outcome},
author = {Kathryn Young-Shand and Patrice C. Roy and Syed Sibte Raza Abidi and Michael Dunbar and Janie Astephen Wilson},
year = {2019},
date = {2019-06-19},
booktitle = {2nd International Combined Meeting of Orthopaedic Research Societies, June 19-22, 2019, Montreal, Canada},
abstract = {Purpose:
Identifying knee osteoarthritis (OA) patient phenotypes is relevant to assessing treatment efficacy, yet biomechanical variability has not been applied to phenotyping. This study aimed to identify demographic and gait related groups (clusters) among total knee arthroplasty (TKA) candidates, and examine inter-cluster differences in gait feature improvement post-TKA.
Method:
Knee OA patients scheduled for TKA underwent three-dimensional gait analysis one-week pre and one-year post-TKA, capturing lower-limb external ground reaction forces and kinematics using a force platform and optoelectronic motion capture. Principal component analysis was applied to frontal and sagittal knee angle and moment waveforms (n=135 pre-TKA, n=106 post-TKA), resulting in a new uncorrelated dataset of subject PCscores and PC vectors, describing major modes of variability throughout one gait cycle (0-100%). Demographics (age, gender, body mass index (BMI), gait speed), and gait angle and moment PCscores were standardized and assessed for outliers. One patient exceeding Tukey’s outer (3*IQR) fence was removed. Two-dimensional multidimensional scaling followed by k-medoids clustering was applied to scaled demographics and pre-TKA PCscores [134x15]. Number of clusters (k=2:10) were assessed by silhouette coefficients, s, and stability by Adjusted Rand Indices (ARI) of 100 data subsets. Clusters were validated by examining inter-cluster differences at baseline, and inter-cluster gait changes (PostPCscore–PrePCscore, n=105) by k-way ANOVA and Tukey's honestly significant difference (HSD) criterion.
Results:
Four (k=4) TKA candidate groups yielded optimum clustering metrics (s=0.4, ARI=0.75). Cluster 1 was all-males (male:female=19:0) who walked with faster gait speeds (1>2,3), larger flexion angle magnitudes and stance-phase angle range (PC1 & PC4 1>2,3,4), and more flexion (PC2 1>2,3,4) and adduction moment (PC2 & PC3 1>2,3) range patterns. Cluster 1 had the most dynamic kinematics and kinetic loading/unloading range amongst the clusters, representing a higher-functioning (less “stiff”) male subset. Cluster 2 captured older (2>1,3) males (31:1) with slower gait speeds (2<1,4), and less flexion moment (PC2 2<1,4) and adduction moment (PC2 2<1,4) range, describing an older, stiff-gait male subset. Cluster 3 was mostly (4:36) females with slower gait speeds (3<1,4), higher BMIs (3>4), and lower flexion angle magnitude (PC1 3<1,2,4) and flexion moment range (PC2 3<1,2,4) features. Cluster 3 showed the “stiffest” gait amongst the clusters, representing a more-obese, stiff-gait female subset. Cluster 4 was mostly (2:41) females with faster gait speeds (4>2,3) and less stiff kinematic and kinetic patterns relative to Clusters 2 and 3, representing a higher-functioning female subset. Radiographic severity did not differ between clusters (Kellgren-Lawrence Grade, p=0.9, n=102), and after removing demographics and re-clustering, gender differences remained (p<0.04). Pre-TKA, higher-functioning clusters (1&4) had more dynamic loading/un-loading kinetic patterns. Post-TKA, high-functioning clusters experienced less gait improvement (flexion angle PC2, 1,4<3, p≥0.004; flexion moment PC2, 4<2,3), with some sagittal range patterns decreasing postoperatively.
Conclusion:
TKA candidates can be characterized by four clusters, differing by demographics and biomechanical severity features. Post-TKA, functional gains were cluster-specific; stiff-gait clusters experienced more improvement, while higher-functioning clusters experienced less gain and showed some decline. Results suggest the presence of cohorts who may not benefit functionally from TKA. Cluster profiling may support triaging and developing targeted OA treatment strategies, meeting individual function needs.},
keywords = {Clustering, Total knee arthroplasty},
pubstate = {published},
tppubtype = {inproceedings}
}
Purpose:
Identifying knee osteoarthritis (OA) patient phenotypes is relevant to assessing treatment efficacy, yet biomechanical variability has not been applied to phenotyping. This study aimed to identify demographic and gait related groups (clusters) among total knee arthroplasty (TKA) candidates, and examine inter-cluster differences in gait feature improvement post-TKA.
Method:
Knee OA patients scheduled for TKA underwent three-dimensional gait analysis one-week pre and one-year post-TKA, capturing lower-limb external ground reaction forces and kinematics using a force platform and optoelectronic motion capture. Principal component analysis was applied to frontal and sagittal knee angle and moment waveforms (n=135 pre-TKA, n=106 post-TKA), resulting in a new uncorrelated dataset of subject PCscores and PC vectors, describing major modes of variability throughout one gait cycle (0-100%). Demographics (age, gender, body mass index (BMI), gait speed), and gait angle and moment PCscores were standardized and assessed for outliers. One patient exceeding Tukey’s outer (3*IQR) fence was removed. Two-dimensional multidimensional scaling followed by k-medoids clustering was applied to scaled demographics and pre-TKA PCscores [134x15]. Number of clusters (k=2:10) were assessed by silhouette coefficients, s, and stability by Adjusted Rand Indices (ARI) of 100 data subsets. Clusters were validated by examining inter-cluster differences at baseline, and inter-cluster gait changes (PostPCscore–PrePCscore, n=105) by k-way ANOVA and Tukey's honestly significant difference (HSD) criterion.
Results:
Four (k=4) TKA candidate groups yielded optimum clustering metrics (s=0.4, ARI=0.75). Cluster 1 was all-males (male:female=19:0) who walked with faster gait speeds (1>2,3), larger flexion angle magnitudes and stance-phase angle range (PC1 & PC4 1>2,3,4), and more flexion (PC2 1>2,3,4) and adduction moment (PC2 & PC3 1>2,3) range patterns. Cluster 1 had the most dynamic kinematics and kinetic loading/unloading range amongst the clusters, representing a higher-functioning (less “stiff”) male subset. Cluster 2 captured older (2>1,3) males (31:1) with slower gait speeds (2<1,4), and less flexion moment (PC2 2<1,4) and adduction moment (PC2 2<1,4) range, describing an older, stiff-gait male subset. Cluster 3 was mostly (4:36) females with slower gait speeds (3<1,4), higher BMIs (3>4), and lower flexion angle magnitude (PC1 3<1,2,4) and flexion moment range (PC2 3<1,2,4) features. Cluster 3 showed the “stiffest” gait amongst the clusters, representing a more-obese, stiff-gait female subset. Cluster 4 was mostly (2:41) females with faster gait speeds (4>2,3) and less stiff kinematic and kinetic patterns relative to Clusters 2 and 3, representing a higher-functioning female subset. Radiographic severity did not differ between clusters (Kellgren-Lawrence Grade, p=0.9, n=102), and after removing demographics and re-clustering, gender differences remained (p<0.04). Pre-TKA, higher-functioning clusters (1&4) had more dynamic loading/un-loading kinetic patterns. Post-TKA, high-functioning clusters experienced less gait improvement (flexion angle PC2, 1,4<3, p≥0.004; flexion moment PC2, 4<2,3), with some sagittal range patterns decreasing postoperatively.
Conclusion:
TKA candidates can be characterized by four clusters, differing by demographics and biomechanical severity features. Post-TKA, functional gains were cluster-specific; stiff-gait clusters experienced more improvement, while higher-functioning clusters experienced less gain and showed some decline. Results suggest the presence of cohorts who may not benefit functionally from TKA. Cluster profiling may support triaging and developing targeted OA treatment strategies, meeting individual function needs.
Identifying knee osteoarthritis (OA) patient phenotypes is relevant to assessing treatment efficacy, yet biomechanical variability has not been applied to phenotyping. This study aimed to identify demographic and gait related groups (clusters) among total knee arthroplasty (TKA) candidates, and examine inter-cluster differences in gait feature improvement post-TKA.
Method:
Knee OA patients scheduled for TKA underwent three-dimensional gait analysis one-week pre and one-year post-TKA, capturing lower-limb external ground reaction forces and kinematics using a force platform and optoelectronic motion capture. Principal component analysis was applied to frontal and sagittal knee angle and moment waveforms (n=135 pre-TKA, n=106 post-TKA), resulting in a new uncorrelated dataset of subject PCscores and PC vectors, describing major modes of variability throughout one gait cycle (0-100%). Demographics (age, gender, body mass index (BMI), gait speed), and gait angle and moment PCscores were standardized and assessed for outliers. One patient exceeding Tukey’s outer (3*IQR) fence was removed. Two-dimensional multidimensional scaling followed by k-medoids clustering was applied to scaled demographics and pre-TKA PCscores [134x15]. Number of clusters (k=2:10) were assessed by silhouette coefficients, s, and stability by Adjusted Rand Indices (ARI) of 100 data subsets. Clusters were validated by examining inter-cluster differences at baseline, and inter-cluster gait changes (PostPCscore–PrePCscore, n=105) by k-way ANOVA and Tukey's honestly significant difference (HSD) criterion.
Results:
Four (k=4) TKA candidate groups yielded optimum clustering metrics (s=0.4, ARI=0.75). Cluster 1 was all-males (male:female=19:0) who walked with faster gait speeds (1>2,3), larger flexion angle magnitudes and stance-phase angle range (PC1 & PC4 1>2,3,4), and more flexion (PC2 1>2,3,4) and adduction moment (PC2 & PC3 1>2,3) range patterns. Cluster 1 had the most dynamic kinematics and kinetic loading/unloading range amongst the clusters, representing a higher-functioning (less “stiff”) male subset. Cluster 2 captured older (2>1,3) males (31:1) with slower gait speeds (2<1,4), and less flexion moment (PC2 2<1,4) and adduction moment (PC2 2<1,4) range, describing an older, stiff-gait male subset. Cluster 3 was mostly (4:36) females with slower gait speeds (3<1,4), higher BMIs (3>4), and lower flexion angle magnitude (PC1 3<1,2,4) and flexion moment range (PC2 3<1,2,4) features. Cluster 3 showed the “stiffest” gait amongst the clusters, representing a more-obese, stiff-gait female subset. Cluster 4 was mostly (2:41) females with faster gait speeds (4>2,3) and less stiff kinematic and kinetic patterns relative to Clusters 2 and 3, representing a higher-functioning female subset. Radiographic severity did not differ between clusters (Kellgren-Lawrence Grade, p=0.9, n=102), and after removing demographics and re-clustering, gender differences remained (p<0.04). Pre-TKA, higher-functioning clusters (1&4) had more dynamic loading/un-loading kinetic patterns. Post-TKA, high-functioning clusters experienced less gait improvement (flexion angle PC2, 1,4<3, p≥0.004; flexion moment PC2, 4<2,3), with some sagittal range patterns decreasing postoperatively.
Conclusion:
TKA candidates can be characterized by four clusters, differing by demographics and biomechanical severity features. Post-TKA, functional gains were cluster-specific; stiff-gait clusters experienced more improvement, while higher-functioning clusters experienced less gain and showed some decline. Results suggest the presence of cohorts who may not benefit functionally from TKA. Cluster profiling may support triaging and developing targeted OA treatment strategies, meeting individual function needs.
Kathryn Young-Shand; Patrice C. Roy; Syed Sibte Raza Abidi; Michael Dunbar; Janie Astephen Wilson
Clinical and Biomechanical Cluster Classification Before TKA Impacts Functional Outcome Proceedings Article
In: Orthopaedic Research Society Conference (ORS), February 2-5, 2019 Austin, Texas, USA, 2019, (Recipient of the American Academy of Orthopaedic Surgeons (AAOS) Women’s Health Advisory Board (WHAB) Award).
BibTeX | Tags: Clustering, Total knee arthroplasty
@inproceedings{Young-Shand2019,
title = {Clinical and Biomechanical Cluster Classification Before TKA Impacts Functional Outcome},
author = {Kathryn Young-Shand and Patrice C. Roy and Syed Sibte Raza Abidi and Michael Dunbar and Janie Astephen Wilson},
year = {2019},
date = {2019-02-02},
booktitle = {Orthopaedic Research Society Conference (ORS), February 2-5, 2019 Austin, Texas, USA},
note = {Recipient of the American Academy of Orthopaedic Surgeons (AAOS) Women’s Health Advisory Board (WHAB) Award},
keywords = {Clustering, Total knee arthroplasty},
pubstate = {published},
tppubtype = {inproceedings}
}