Tracking Foot Drop Recovery Following Lumbar-Spine Surgery, Applying Multiclass Gait Classification Using Machine...

Area of Science:

• Orthopedics
• Biomechanical Engineering
• Machine Learning in Healthcare

Background:

• Accurate human gait evaluation is crucial for orthopedic foot and ankle surgeons to monitor patient recovery.
• Current visual gait inspection methods are subjective and lack accuracy, hindering optimal diagnosis and treatment.
• Objective gait assessment can significantly improve patient outcomes in orthopedic care.

Purpose of the Study:

• To develop an accurate and clinically applicable method for evaluating the foot drop condition.
• To utilize machine learning for objective gait analysis in patients undergoing surgical treatment for foot drop.
• To identify the most effective machine learning algorithm for categorizing foot drop recovery stages.

Main Methods:

• Gait data were collected from 56 patients with L5-origin foot drop using inertial measurement unit sensors.
• Machine learning algorithms were applied to analyze gait data across different surgical treatment recovery stages.
• The random forest algorithm was selected for its superior performance in classification and regression tasks.

Main Results:

• The random forest algorithm demonstrated the highest classification accuracy at 84.89% among the tested machine learning models.
• The random forest algorithm achieved a mean absolute error of 0.3785 for regression, indicating precise outcome prediction.
• Machine learning effectively categorized patient gait data according to recovery stages.

Conclusions:

• Machine learning, particularly the random forest algorithm, offers a highly accurate and objective approach to foot drop gait evaluation.
• This objective assessment method can enhance the tracking of patient recovery and inform treatment adjustments.
• The findings support the clinical integration of machine learning for improved orthopedic patient management and outcomes.