Diagnosis of Mechanical Low Back Pain Using a Fuzzy Logic-Based Approach

Abstract

Abstract: Back pain is one of the main causes of disability, and its proper diagnosis and treatment are difficult tasks. Intelligent methods can help physicians make a more precise diagnosis of diseases. The present study was conducted to diagnose the correct type of mechanical low back pain (LBP) using an Adaptive Neuro-Fuzzy Inference System (ANFIS). The diagnostic parameters of mechanical LBP were determined using library reviews and the views of experts based on the Delphi technique. Modelling was done in MATLAB R2012 using the ANFIS. After the modelling stage, the method was tested in terms of the percentage of corrected classification and diagnostic value indicators. Modelling is applied in the present study to diagnose different types of mechanical LBP, including back strain, spondylolisthesis, spinal stenosis, disc herniation, and scoliosis. The modelling input included 17 diagnostic parameters, and its output contained various types of mechanical LBP. The percentage of corrected classification varied from 80.9% to 83.8% (disc herniation and spondylolisthesis). The system test in the present study showed an appropriate accuracy in diagnosing different types of mechanical LBP. As a result, this system can be helpful in clinical settings for diagnosing different types of mechanical LBP presenting with similar symptoms.