Skills:
programming(Python, MATLAB);
communication skills;
research paper writing;
modeling and data visualization
Influence of Experience on Eye Gaze Patterns and Identification of Normative Gait from Biological Motion
I-Chieh Lee, Matheus Maia Pacheco, Ziwei Liu, He Helen Huang.
ICOPA International Conference On Patient Advocacy, 2019
Motivation
Gait dysfunctions often lead to walking patterns that are different from able-bodied walking patterns. Physical therapists (PT) usually rely on observations to diagnose and plan interventions specific to these patterns.
It is expected that PT students learn to extract features from these patterns through education and practices. So, where do they look at to evaluate gait patterns? Could we use this information to determine whether the diagnosis is reliable or how they transform knowledge into practices?
We compared what body areas PT students and novices are looking at for gathering information when evaluating gait patterns. We used principal component analysis to transform joint motion of two amputee walkers and a able-bodied walkers into three biological motion models. We then build six videos from each model displaying a synthesized point-light walker for participants to observe. They were asked to rank the videos based on how normal they appeared.
Contribution
In this project, I participated in data modeling and user studies. I learned the basics of gait cycles and principal component analysis through practice.
I recruited ten subjects and conducted user studies with them. Additionally, I was responsible for data cleaning and processing. I explored different ways of analyzing the heat map of participants' fixations, including basic machine learning and computer vision techniques.
We found that novices spent most time fixating the legs (78.37%) while PT students spent relatively more time on pelvic (21%) and shoulder (20.93%). There were also follow-up studies to investigate this direction and prepare domain knowledge for advanced machine learning projects.
And beyond...
Can you tell these dots of motion (biological motion) represent human walking? Can you rate the gait pattern (1-6) in the video? Wherein, 6 represents the most normal gait pattern. The answer is at the end of the video. Please check out my former colleague's recent paper related to this topic (Perceiving amputee gait from biological motion: kinematics cues and effect of experience level).
I joined this project, wondering how do PTs handle different cases in their work. Furthermore, I am still curious about using their domain knowledge to make their job easier by introducing new technology such as computer vision. Or, would they want that?