Advanced Machine Learning Models for Automating Lower Back Pain Diagnosis Using Biomechanical Data

Lower back pain (LBP) is a common health issue with significant societal impact. Traditional diagnosis relies on MRI, subjective assessments and clinical expertise, often causing delays in accurate detection. This study explores machine learning (ML) techniques to automate LBP diagnosis using biomechanical data. Exploratory data analysis, including correlation heatmaps, identified key features such as pelvic tilt, sacral slope, and lumbar lordosis angle associated with LBP. Various ML models were applied, including Random Forest, Support Vector Machines, Naïve Bayes, and ensemble methods like Gradient Boosting, AdaBoost, CatBoost, LightGBM, and XGBoost. Among them, the Random Forest classifier achieved the highest accuracy of 90%. It yielded a precision of 0.83, recall of 0.95, and F1-score of 0.86 for normal pain, and a precision of 1.0, recall of 0.82, and F1-score of 0.90 for abnormal pain. The results demonstrate the potential of ML techniques to provide cost-effective, scalable and accurate LBP diagnostics.