Abstract / Summary
Thermal imaging provides a non-contact approach for continuous environmental monitoring in intensive care units, although most existing studies remain limited to controlled or experimental settings with limited real-world validation. This study aimed to design, deploy, and operationally evaluate a scalable, low-cost, on-premise thermal Internet of Medical Things system in a real intensive care unit. A distributed network of seven thermal acquisition nodes, scalable to thirteen beds, was implemented using embedded devices and radiometric infrared cameras. Thermal images were transmitted through the hospital Wi-Fi network to a local server for storage, indexing, and near real-time visualization. System performance was evaluated using operational metrics including end-to-end latency, throughput, availability, and connectivity incidents. During a 10-day structured evaluation, the system generated 5,763,312 thermal images (approximately 1.47 TB), achieving an aggregate acquisition rate of 6.67 images per second. After architectural optimization, mean latency decreased from 26,344 ms to 126 ms (99.52% reduction), with a median of 117 ms and a 95th percentile of 207 ms. Effective system availability reached 95.29%, with most incidents associated with transient wireless connectivity interruptions. These results demonstrate the feasibility of deploying a distributed thermal IoMT system in a real intensive care unit under hospital information technology governance, offering a preliminary reference model for on-premise, nonintrusive thermal sensing infrastructures in critical care, pending validation across additional institutional and hardware settings.