A novel scoring approach for the Wolf Motor Function Test in stroke survivors using motion-sensing technology and...

Area of Science:

• Rehabilitation technology
• Biomedical engineering
• Neuroscience

Background:

• Clinical scales for upper-limb motor function assessment, like the Wolf Motor Function Test-Functional Ability Scale (WMFT-FAS), are subjective and labor-intensive.
• Objective and efficient methods are needed to evaluate motor function in stroke survivors.

Purpose of the Study:

• To propose and validate a novel, objective scoring approach for motor function assessment in stroke survivors.
• To overcome the limitations of traditional subjective clinical scales.

Main Methods:

• Utilized a Microsoft Kinect v2 for capturing functional movements.
• Developed a customized motion tracking system for data recording.
• Implemented an intelligent scoring system with a feed-forward neural network (FFNN) for movement quality evaluation and WMFT-FAS scoring.
• Applied singular spectrum analysis and multi-ReliefF for enhanced usability.

Main Results:

• Achieved high accuracy (0.924 ± 0.027) and F1-score (0.875 ± 0.063) in validation against a physiotherapist's scores.
• Demonstrated strong performance with sensitivity (0.915 ± 0.051), specificity (0.969 ± 0.013), and AUC (0.952 ± 0.038).
• Reported low mean absolute error (0.098 ± 0.037) and root mean squared error (0.214 ± 0.078).

Conclusions:

• The proposed novel scoring approach provides objective and accurate motor function assessment.
• This technology can assist physiotherapists in making individualized treatment decisions for stroke survivors.