An Algorithm for Accurate Marker-Based Gait Event Detection in Healthy and Pathological Populations During Complex...

Area of Science:

• Biomechanics and Movement Science
• Medical Technology and Instrumentation
• Rehabilitation Engineering

Background:

• Quantifying gait in complex motor tasks requires accurate gait event (GE) detection beyond straight-line walking.
• Existing methods face challenges in non-straight walking and diverse populations, limiting comprehensive gait analysis.

Purpose of the Study:

• To propose and validate a novel marker-based GE detection method suitable for curvilinear walking and step negotiation.
• To assess the method's performance against existing algorithms and its impact on stride parameter calculations.
• To evaluate the method's efficacy across healthy young adults and individuals with various gait impairments.

Main Methods:

• A new marker-based GE detection algorithm was developed and tested.
• The method was validated against a pressure insole-based reference system.
• Performance metrics including sensitivity, PPV, F1-score, bias, precision, and accuracy were calculated.
• The study included healthy young adults and five clinical cohorts (older adults, COPD, MS, PD, hip fracture).

Main Results:

• The proposed method achieved high accuracy (≥99% sensitivity, PPV, F1-score) in young adults, outperforming existing algorithms with minimal bias (<10 ms).
• Temporal inaccuracies from GE detection had a minor impact on stride parameters (median absolute errors ≤1%).
• Similar high performance and excellent agreement (ICC) with pressure insoles were observed across all clinical cohorts and tasks.

Conclusions:

• The developed marker-based method accurately detects gait events under diverse walking conditions and for various gait impairments.
• This method offers a reliable tool for quantitative gait analysis in complex tasks and clinical populations.
• The findings support the use of this advanced technique for improved gait assessment and rehabilitation strategies.