A Multi-Level Sensor-Based Spinal Cord Disorder Classification Model for Patient Wellness and Remote Monitoring

Detection of sagittal spine and spinopelvic disorder or injury is a key issue in children and adults due to variation in shape and orientation. Especially in remote patient monitoring for proper diagnose the structure and angular deviation of the lumbar spine and pelvis are required. So automatic rehabilitation process, especially for bedridden people at home, is very much essential for spinal cord injury patients to optimize the quick diagnosis process. The main objective of this work is to improve the classification rate of the vertebral spinal cord sensing data for automatic rehabilitation process. In this paper, a filter based multi-level segmentation and classification approach is implemented on the vertebral column dataset. In this approach, sensor generated spinal cord data are used in order to predict the severity level of each spinal cord disorder. In this approach, a novel vertebral data pre-processing method, a multi-level sensing approach and an improved random forest technique is proposed to predict the disorder with high true positive rate. Experimental results proved that the present model has better efficiency than the existing vertebral classifiers such as Naïve Bayes, Neural Network, Adaboost, Random forest and SVM in terms of true positive rate; TP = 0.9813, Accuracy = 0.9783, Error rate = 0.0246 are concerned.