Predictive System for Risk of Herniated Disc via Smartphone using Data Mining Techniques

Abstract

This research proposed a prediction system for the risk of herniated disc using data mining techniques on smartphone. A total of 500 data sets were collected and used for sample data. We applied three data mining techniques, namely, Decision Tree, Artificial Neural Network, and Naïve Bay to create three prediction system models.  Then, the levels of efficiency of the three models were measured and compared in terms of accuracy value, precision value, recall value, and RMSE value.  After that, the models were analyzed to identify the factors that affected the risk of the occurring of herniated disc and then the model with the highest efficiency was chosen for application development.  The results revealed that (1) the most efficient model for prediction of herniated disc was the prediction model based on Artificial Neural Network technique with an accuracy of 94.60%, precision of 94.60%, recall of 94.60% and RMSE of 0.218; (2) the result of identifying the factors affecting the herniated disc revealed that the most important factor was the pain in the hips, soles or toes, to be followed by the frequent lying prone down, and the condition of weakening legs, respectively; and (3) the result of evaluation of the users’ satisfaction with the developed system showed that the rating mean for users’ satisfaction was 4.20 with standard deviation of 0.53, indicating that the users were satisfied with the system at the high level.  In conclusion, the developed system is appropriate and can be applied for analysis of the risk of herniated disc.  It helps the patients to analyze the risk of herniated disc on their own and can suggest guidelines for preliminary treatment of the disease.