Grouping Runners Based on Three-Dimensional Ankle Joint Angles Using a Deep Temporal Clustering Algorithm
Gespeichert in:
| Titel: | Grouping Runners Based on Three-Dimensional Ankle Joint Angles Using a Deep Temporal Clustering Algorithm |
|---|---|
| Autoren: | Zaniar Mohamadi, Mansour Eslami, Rohollah Yousefpour |
| Quelle: | Scientific Journal of Rehabilitation Medicine. 14:418-433 |
| Verlagsinformationen: | Negah Scientific Publisher, 2025. |
| Publikationsjahr: | 2025 |
| Beschreibung: | Background and Aims Static classification methods have been used to investigate the relationship between foot type and structure with risk factors and to prescribe preventive and therapeutic interventions such as orthotics and shoes. However, a weak relationship between static measurements and foot movement performance has been reported, and studies have shifted towards using dynamic foot classification methods based on movement patterns. Therefore, this study aimed to group runners based on three-dimensional kinematic patterns of the ankle joint during running and to investigate significant differences in kinematic patterns among the identified groups. Methods The deep temporal clustering algorithm was implemented during barefoot running to identify homogeneous subgroups on three-dimensional ankle joint angle data of 108 healthy adults (age: 22.45±2.42 years; height: 1.69±0.11 m; body mass: 64.66±9.54 kg; gender: 55 males, 53 females). After identifying the clusters, a parametric statistical mapping was used to examine the differences in ankle joint kinematic patterns across the stance phase of running in the identified clusters. Results Three distinct subgroups were identified. A comparison of the time series curves showed that individuals in cluster 1 had a larger average ankle joint dorsiflexion range compared to the other two cluster patterns between 40% and 80% of the stance phase of running (P=0.004). Additionally, individuals in cluster 3 showed a greater range of ankle joint eversion between 60% and 100% of the stance phase of running compared to the patterns of individuals in the other two clusters (P=0.038). However, changes in ankle joint angles in the horizontal plane during running were similar in all three groups. Conclusion The proposed model can automatically group runners based on the kinematic pattern of the ankle joint during running. By identifying the movement pattern of homogeneous groups and determining appropriate interventions for each of these groups, a suitable guideline for prescribing preventive interventions such as shoes can be developed. |
| Publikationsart: | Article |
| ISSN: | 2252-0414 2251-8401 |
| DOI: | 10.32598/sjrm.14.3.3334 |
| Dokumentencode: | edsair.doi...........5a07177f4340ca46d1e6f77bb4d9b2f8 |
| Datenbank: | OpenAIRE |
Nájsť tento článok vo Web of Science | Abstract: | Background and Aims Static classification methods have been used to investigate the relationship between foot type and structure with risk factors and to prescribe preventive and therapeutic interventions such as orthotics and shoes. However, a weak relationship between static measurements and foot movement performance has been reported, and studies have shifted towards using dynamic foot classification methods based on movement patterns. Therefore, this study aimed to group runners based on three-dimensional kinematic patterns of the ankle joint during running and to investigate significant differences in kinematic patterns among the identified groups. Methods The deep temporal clustering algorithm was implemented during barefoot running to identify homogeneous subgroups on three-dimensional ankle joint angle data of 108 healthy adults (age: 22.45±2.42 years; height: 1.69±0.11 m; body mass: 64.66±9.54 kg; gender: 55 males, 53 females). After identifying the clusters, a parametric statistical mapping was used to examine the differences in ankle joint kinematic patterns across the stance phase of running in the identified clusters. Results Three distinct subgroups were identified. A comparison of the time series curves showed that individuals in cluster 1 had a larger average ankle joint dorsiflexion range compared to the other two cluster patterns between 40% and 80% of the stance phase of running (P=0.004). Additionally, individuals in cluster 3 showed a greater range of ankle joint eversion between 60% and 100% of the stance phase of running compared to the patterns of individuals in the other two clusters (P=0.038). However, changes in ankle joint angles in the horizontal plane during running were similar in all three groups. Conclusion The proposed model can automatically group runners based on the kinematic pattern of the ankle joint during running. By identifying the movement pattern of homogeneous groups and determining appropriate interventions for each of these groups, a suitable guideline for prescribing preventive interventions such as shoes can be developed. |
|---|---|
| ISSN: | 22520414 22518401 |
| DOI: | 10.32598/sjrm.14.3.3334 |