Multidimensional machine learning approach for classifying patients with neck pain based on movement control test.

Area of Science:

• Biomechanical analysis
• Rehabilitation technology
• Clinical assessment tools

Background:

• Neck pain (NP) is frequently associated with impaired cervical movement control.
• Advancements in movement control assessment technology allow for detailed parameter extraction.
• The potential for precise classification using these parameters remains underexplored.

Purpose of the Study:

• To evaluate the validity of various parameters and difficulty levels of the Butterfly cervical movement control test.
• To assess the classification performance of different machine learning models using these parameters.

Main Methods:

• Sixty-five NP patients and fifty asymptomatic controls performed the Butterfly test.
• Machine learning models (Random Forest, SVM, Logistic Regression) were built using individual or multiple parameters across difficulty levels.
• Performance metrics included AUC, accuracy, F1 score, precision, recall, specificity, information gain, and Gini index.

Main Results:

• Models utilizing amplitude accuracy, time-on-target, or undershoot demonstrated high classification performance.
• Combining directional accuracy or all parameters across all difficulty levels resulted in the most balanced classification probability.

Conclusions:

• Directional accuracy alone or all parameters of the Butterfly test, across all difficulty levels, are recommended for classifying NP patients.
• This approach provides a clinically valid tool for assessing neck pain patients.