Better understanding fall risk: AI-based computer vision for contextual gait assessment.

Area of Science:

• Neurology
• Biomedical Engineering
• Computer Science

Background:

• Parkinson's disease (PD) increases fall risk, necessitating accurate assessment methods.
• Wearable inertial-based measurement units (IMUs) capture gait dynamics but miss environmental context.
• Current fall risk assessment in PD often overlooks extrinsic factors like obstacles.

Purpose of the Study:

• To integrate wearable video-based eye-tracking with AI computer vision for enhanced fall risk assessment in Parkinson's disease.
• To combine intrinsic gait data from IMUs with extrinsic environmental data in free-living settings.
• To develop a more comprehensive approach to understanding fall risk in Parkinson's disease patients at home.

Main Methods:

• Utilized ergonomic wearable video-based eye-tracking glasses.
• Employed AI-based computer vision methodologies for data analysis.
• Collected data in free-living, home-based environments from a small group of people with PD.

Main Results:

• Demonstrated the feasibility of using video-based eye-tracking and AI in home environments for PD research.
• Showcased the potential to complement IMU data with environmental awareness.
• Provided a foundation for more holistic fall risk assessment in Parkinson's disease.

Conclusions:

• Video-based eye-tracking coupled with AI offers an ethical and efficient method to capture environmental context for fall risk assessment in PD.
• This integrated approach represents an evolutionary step beyond IMU-only methods for Parkinson's disease research.
• Future research can leverage this technology for a more comprehensive understanding and mitigation of fall risks in PD.