Detection of Autonomic Dysreflexia in Persons with Cervical Spinal Cord Injuries through a Support Vector Machine

One of the more common post-SCI health conditions is Autonomic Dysreflexia (AD) and poor management of AD can be life threatening. Key physiological indicators of AD include an altered heart rate and pathological sweating above the injury site. Utilizing commercially-available galvanic skin resistance (GSR), heart rate, and skin temperature sensors, we developed a mobile computer application to enable wearers to monitor these physiological indicators in real-time and throughout the day. Subjects were recruited to regularly use the system for two weeks and report any incidences of AD. A Support Vector Machine was used to develop a machine learning model that could predict the occurrence of AD with a high level of accuracy. The objective of this study is to provide a generalizable recognition system to reliably detect the onset of AD in the vast majority of tetraplegics without prior training. Continuous physiological monitoring to alert tetraplegics and their caregivers of the onset of AD would promote greater independence and community participation.