Power wheelchair driving analysis for people with motor disabilities using ANN classification

Area of Science:

• Rehabilitation Engineering
• Artificial Intelligence in Healthcare
• Biomedical Signal Processing

Background:

• Assessing the driving skills of patients with motor deficiencies using electric wheelchairs is crucial for safety.
• Traditional methods may not objectively evaluate the complex driving abilities required for power wheelchair (PW) navigation.
• Severe motor deficiencies can make operating a PW dangerous, necessitating advanced assessment tools.

Purpose of the Study:

• To analyze the driving skills of patients with various pathologies using artificial neural networks (ANNs).
• To objectively evaluate the safety and efficacy of power wheelchair use in patients with motor impairments.
• To develop a model for classifying driver skill levels (familiarized vs. novice) based on simulated driving performance.

Main Methods:

• An experimental study utilizing a Power Wheelchair (PW) driving simulator was conducted.
• Artificial neural networks were employed to analyze patient driving trajectories and identify errors.
• The error between reference and calculated trajectories served as input for a classification model.

Main Results:

• The developed model successfully classified users as familiarized or novice based on driving error patterns.
• The evolution of trajectory error was identified as a key indicator of driving skill improvement during the learning phase.
• Objective data on driving performance was gathered, highlighting potential risks associated with severe motor deficiencies.

Conclusions:

• Artificial neural networks and PW driving simulators provide an objective method for assessing patient driving skills.
• The trajectory error analysis model can effectively evaluate and guide the improvement of driving abilities.
• This approach enhances patient safety and optimizes rehabilitation strategies for power wheelchair users.