Novel methodology for estimating Initial Contact events from accelerometers positioned at different body locations

Area of Science:

• Biomechanics
• Signal Processing
• Wearable Technology

Background:

• Initial Contact events (ICE) are crucial for gait cycle segmentation and parameter calculation.
• Current algorithms often require specific accelerometer placement, limiting long-term and real-world applications.
• Robustness to sensor positioning variations is needed for widespread accelerometry adoption in daily living.

Purpose of the Study:

• To develop a novel methodology for estimating ICE from accelerometers placed at diverse body locations.
• To address the need for algorithms adaptable to sensor placement changes in long-term gait analysis.
• To enhance the reliability of gait analysis in real-life settings.

Main Methods:

• The proposed DK-TiFA methodology leverages domain knowledge of spectral relationships between body part movements during gait.
• Time-frequency analysis of acceleration signals is driven by these fundamental spectral relationships.
• The approach was validated on four large public gait databases encompassing 613 subjects and 7 sensor locations.

Main Results:

• DK-TiFA demonstrated high accuracy and robustness in estimating ICE across various accelerometer specifications.
• The methodology proved effective despite variations in gait speeds and environmental conditions.
• Performance was consistently high across multiple body locations, including ankle, thigh, waist, chest, and arm.

Conclusions:

• The DK-TiFA methodology offers a robust solution for accurate ICE detection from multi-location accelerometers.
• This approach enhances the feasibility of long-term, real-world gait analysis with wearable sensors.
• DK-TiFA's adaptability to sensor placement and environmental factors makes it suitable for diverse clinical and research applications.