Integrating Multidimensional Data Analytics for Precision Diagnosis of Chronic Low Back Pain

Low back pain (LBP) is a leading cause of disability worldwide, with up to 25% of cases become chronic (cLBP). Optimal diagnostic tools for cLBP remains unclear. Here we leveraged a comprehensive multi-dimensional data-set and machine learning-based feature importance selection to identify the most effective diagnostic tools for cLBP patient stratification. The dataset included questionnaire data, clinical and functional assessments, and spino-pelvic magnetic resonance imaging (MRI), encompassing a total of 144 parameters from 1,161 adults with (n=512) and without cLBP (n=649). Boruta and random forest were utilised for variable importance selection and cLBP classification respectively. Boruta feature selection led to pronounced variable reduction (median of all 15 datasets: 63.3%), while performing comparable to using all variables across all modality datasets. Multi-modality models performed better than single modality models. Boruta selected key variables from questionnaire, clinical, and MRI data were the most effective in distinguishing cLBP patients from controls with an AUC (area under the receiver operating characteristic curve) of 0.699 (95% confidence interval [CI], 0.669 – 0.729). The most robust features (n=9) across the whole dataset identified were psychosocial factors, neck and hip mobility, as well as lower lumbar disc herniation and degeneration. These critical variables (AUC = 0.664, 95% CI = 0.514 – 0.814) outperformed all parameters (AUC = 0.602, 95% CI = 0.538 – 0.666) in an unseen holdout dataset, demonstrating superior patient delineation. Paving the way for targeted diagnosis and personalized treatment strategies, ultimately enhancing clinical outcomes for cLBP patients.