Research Abstracts (Poster)
08:30 - 19:30 | Wed 26 Oct | Auditorium Foyer | WePOS
11:45 - 12:15 | Thu 27 Oct | Main Auditorium | IS-2
Background: Many automated, non-contact cardiac pulse (heart rate and heart rate variability) measurement systems have been developed in the optical and biomedical signal processing communities [1,2,3,4]. However, the success of these applications is limited due to their complex processing requirements (offline or using blind source separation techniques) and potential for optical occlusion. Examples of real-time situations where automated, non-contact monitoring of multiple subjects is useful includes hospital emergency department waiting rooms, educational classrooms, and senior homes. These environments require scalable systems that can monitor as many people as occupancy allows and require real-time feedback for healthcare professionals or local responders. Purpose: To determine the capability of monitoring the cardiac health of multiple subjects without contact and alerting nearby responders of cardiac emergencies or irregularities of subjects, including cardiac arrest, in real-time. Methods: To address the problem of measuring the cardiac pulse of multiple subjects in real-time environments, we have developed a real-time Internet of Things (IoT) system that uses wireless cameras to obtain the cardiac pulse of multiple subjects. All subjects are volunteering college students. A university classroom (50 feet x 60 feet) was used to simulate the use environment. A sensor, aggregator, and human-machine interface architecture were used, where the sensors are wireless cameras, the aggregator is a powerful machine connected to the wireless network, and the human-machine interface is a web application accessible by nearby responders. The experimental IoT system was developed using off-the-shelf hardware (Raspberry Pi, Pi Camera module, and x86 PC). Multiple Raspberry Pi-based cameras are used to record video (1920 x 1080 pixel resolution at 30 Hz) of the sitting subjects, which is sent to the aggregating server on the x86 PC. Once faces in the video are recognized and identified, the cardiac pulse signal is extracted and merged for each detected subject in the videos. The pulses and images are then sent to the web client for monitoring, and in the event of an extreme pulse signal, the client flashes alerts. The primary study outcome measurement was the cardiac pulse (in beats per minute), which was compared to baseline resting pulse measurements from ECGs recorded with an off-the-shelf ECG circuit. Results: Results have been obtained from preliminary experiments of videos taken from two wireless cameras recording groups of four subjects sitting in the classroom simulation. No false positives or false negatives (either false subjects or false cardiac emergencies) have occurred in the samples so far. The next step following these preliminary experiments is to test the robustness of the system with significantly more subjects (as in a large lecture hall) to determine the resolution needed before false positives or negatives are introduced. Conclusions: The implication of this study is the significance of IoT systems in wireless health monitoring. With this type of system, it is possible to effectively monitor multiple subjects in multiple environments while decisively acting in emergencies. Given the portability of the system, aggregated cardiac data can lead to other significant conclusions, such as understanding the general cardiac health of the monitored audience.
No information added