Scholarlink

Gait analysis, which involves the examination of human locomotion, provides valuable insights into underlying disorders and impairments. Recent advancements in artificial intelligence (AI) algorithms have led to the emergence of automated analytical methods for gait analysis. This study focuses on the application of AI-driven analysis of ground reaction force (GRF) patterns to distinguish between individuals with normal gait and those presenting gait disorders. The study utilizes the publicly available GaitRec dataset, which comprises extensive and fully annotated bilateral GRF measurements obtained during walking trials from individuals with diverse musculoskeletal conditions and unaffected individuals serving as controls. The Explainable Boosting Machines (EBM) algorithm was employed to build a binary classifier, named HC-GD (HC: Healthy Controls vs individuals with diverse musculoskeletal conditions (GD: Gait Disorder)). The EBM has been chosen due to its ability to provide both global and local explanations, allowing us to understand the contribution of each feature to the final prediction. The global explanation revealed that the right vertical GRF feature during frames 78–84 played a crucial role in distinguishing between healthy individuals and those with gait disorders. Additionally, the right medio-lateral frames 94–96 also contributed to the classification accuracy. The local explanation confirmed the importance of the last frames of the right vertical GRF for the correct discrimination between HC and GD subjects, but also for the misclassification. The proposed AI-driven analysis of GRF patterns using the GaitRec dataset holds promise for accurate classification of individuals with gait disorders. This research has the potential to revolutionize the field of gait analysis by providing clinicians with a reliable and efficient tool for identifying and classifying gait abnormalities. The findings of this study contribute to the growing body of knowledge in the field of AI-driven gait analysis and pave the way for further advancements in this emerging area of study.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Bringing Explainability to the Prediction of Gait Disorders from Ground Reaction Force (GRF): a Machine Learning Study on GaitRec Dataset

  • 章节

  • 发表日期：

    2026年

查看全文
  • Vera Gramigna,
  • Alessia Sarica

摘要

Gait analysis, which involves the examination of human locomotion, provides valuable insights into underlying disorders and impairments. Recent advancements in artificial intelligence (AI) algorithms have led to the emergence of automated analytical methods for gait analysis. This study focuses on the application of AI-driven analysis of ground reaction force (GRF) patterns to distinguish between individuals with normal gait and those presenting gait disorders. The study utilizes the publicly available GaitRec dataset, which comprises extensive and fully annotated bilateral GRF measurements obtained during walking trials from individuals with diverse musculoskeletal conditions and unaffected individuals serving as controls. The Explainable Boosting Machines (EBM) algorithm was employed to build a binary classifier, named HC-GD (HC: Healthy Controls vs individuals with diverse musculoskeletal conditions (GD: Gait Disorder)). The EBM has been chosen due to its ability to provide both global and local explanations, allowing us to understand the contribution of each feature to the final prediction. The global explanation revealed that the right vertical GRF feature during frames 78–84 played a crucial role in distinguishing between healthy individuals and those with gait disorders. Additionally, the right medio-lateral frames 94–96 also contributed to the classification accuracy. The local explanation confirmed the importance of the last frames of the right vertical GRF for the correct discrimination between HC and GD subjects, but also for the misclassification. The proposed AI-driven analysis of GRF patterns using the GaitRec dataset holds promise for accurate classification of individuals with gait disorders. This research has the potential to revolutionize the field of gait analysis by providing clinicians with a reliable and efficient tool for identifying and classifying gait abnormalities. The findings of this study contribute to the growing body of knowledge in the field of AI-driven gait analysis and pave the way for further advancements in this emerging area of study.