Initial classification of low back and leg pain based on objective functional testing: a pilot study of machine...

Area of Science:

• Orthopedics
• Rehabilitation Medicine
• Biomedical Engineering

Background:

• The five-repetition sit-to-stand (5R-STS) test assesses functional impairment but its clinical interpretability needs further understanding.
• Differences in 5R-STS performance among patients with lumbar disk herniation (LDH), lumbar spinal stenosis (LSS), and chronic low back pain (CLBP) have been observed.

Purpose of the Study:

• To evaluate the diagnostic information provided by the 5R-STS test for differentiating causes of low back pain.
• To assess the accuracy of a machine learning algorithm using 5R-STS data for classifying spinal conditions.

Main Methods:

• Patients were categorized into LDH, LSS, or CLBP groups based on clinical assessment and imaging.
• 5R-STS test times were compared across groups, with adjustments for demographic factors.
• A machine learning model was trained using 5R-STS time, age, gender, height, and weight for classification.

Main Results:

• Significant differences in 5R-STS performance were found among the diagnostic groups.
• The machine learning model achieved 96.2% classification accuracy in internal validation.
• High sensitivity and specificity were reported for classifying LDH, LSS, and CLBP.

Conclusions:

• 5R-STS test performance varies by the underlying cause of back and leg pain, even after accounting for demographics.
• Combined with machine learning, the 5R-STS test can infer the etiology of spinal pain with high accuracy.