Using human-in-the-loop optimization for guiding manual prosthesis adjustments: a proof-of-concept study.

Area of Science:

• Biomedical Engineering
• Rehabilitation Robotics
• Human-Computer Interaction

Background:

• Human-in-the-loop (HITL) optimization is effective for complex interactive problems, including reducing metabolic cost with robotic exoskeletons.
• Many passive prostheses and orthoses require manual setting adjustments, presenting a challenge for optimal user fit and function.

Purpose of the Study:

• To investigate if HITL algorithms can accelerate manual adjustments in a prosthesis fitting simulation.
• To determine the efficiency and accuracy of HITL algorithms in optimizing prosthesis simulator settings for reduced contralateral limb loading rate.

Main Methods:

• Eight healthy participants used a prosthesis simulator, walking at 0.8 m/s under 16 combinations of shoe heel and pylon height.
• A HITL optimization algorithm was employed to identify optimal settings by minimizing the loading rate on the contralateral limb.
• Algorithm performance was assessed using convergence criteria and accuracy compared against a full parameter sweep.

Main Results:

• The HITL optimization successfully reduced the time to find optimal settings in five out of eight participants.
• In three participants, the algorithm either converged late or failed to converge within the study's iterations.
• Accuracy was evaluated against the optimum derived from testing all possible parameter combinations.

Conclusions:

• The HITL methodology shows potential for optimizing manually adjustable assistive devices, such as unpowered prostheses.
• Further research is required to enhance algorithm robustness and assess its applicability with actual prostheses in amputee populations.