Evaluation of the effectiveness of non-contact respiration rate detection for post-crash care application

The ability of emergency medical services (EMS) personnel to understand the mortality and morbidity risks associated with different injuries is critical for effective motor vehicle collision (MVC) post-crash care. Interwoven with this ability is the practice of accurate and efficient patient triage. Triage allows EMS to gain a better understanding of the physiological and physical status of a patient and their injuries, with the intention of identifying additional resources needed, potential treatment options, and the most appropriate care destination. An interactive, in-vehicle triage system, based on the Simple Triage and Rapid Treatment protocol, under development could lead to improved MVC triage accuracy and efficiency. As a component of that system, this study examines the effectiveness of using non-contact respiration rate detection technologies integrated into a vehicle cabin. The selected technologies included ultrasonic respiration detection, thermal respiration detection, and pressurebased detection which were all compared against data collected from a respiration belt. All technologies were integrated into a 2015 Ford Taurus. Testing took place inside the running vehicle and considered occupant weight, cabin temperature, occupant clothing weight, and environmental sound as experimental factors. Respiration rate was then calculated using a 30-second sliding window where the means and standard deviations were used to compare the accuracy and precision of the sensor systems across the experimental factors. The sensor type, temperature level, and sound level main effects significantly influenced the mean respiration rate. For the standard deviation of respiration rate, significant effects were found for the sensor type, sound level, and clothing weight main effects. Among the tested sensor systems, the pressure sensor was found to best match the accuracy and precision of the respiration belt. Future work should address limitations in the approach used for this proof-of-concept work.