Classification and correlational analysis on lower spine parameters using data mining techniques

Abstract

The application of data mining in the field of medical science is slowly gaining popularity. This is due to the fact that enormous statistical inferences from data related to the human body and medicine was a possible with high accuracy rates which was a tedious task in the past. This had led to discoveries and breakthroughs which has saved thousands of lives. Lower back pain is one of the most common issues faced by majority of the population throughout the world. The early detection and treatment of LBP can avoid life threatening issues in the body. Objective: This study aims to create a classification model which can be used to detect an unhealthy spine using the lumbar and sacral parameters. Correlational analysis was performed between different attributes to find distinguishing factors between healthy and unhealthy spine. Method: Classification methods were used such as decision tree and SVM. Correlational analysis was performed using pearson method between each attribute. Results: After creating the model using the different classification methods it was found that Ctree produced the highest accuracy with 92.80% on average. It was also found that there were 6 attribute pairs that had high correlation coefficient to distinguish unhealthy and healthy spine observations.