Accurate Temperature Measurements for Medical Research using Body Sensor Networks

Medical measurements and clinical trials are often carried out in controlled lab settings - severely limiting the realism and duration of such studies. Our goal is henceforth to design a body sensor network for unobtrusive and highly accurate profiling of body parameters over weeks in realistic environments. One example application is monitoring the impact of sleep deprivation on periodic processes in the human body known as circadian rhythms, which requires highly accurate profiling of skin temperature across the human body over weeks with real-time feedback to a remote medic. We analyze the requirements on a body sensor network for such applications and highlight the need for self-organizing behavior such as adaptive sampling to ensure energy efficiency and thus longevity, adaptive communication strategies, self-testing, automatic compensation for environmental conditions, or automatic recording of a diary of activities. As a first step towards this goal, we design and build a prototype of such a non-invasive wearable wireless monitoring system for accurate body temperature measurements and real-time feedback to the medic. Through the design, parameterization, and calibration of an active measurement subsystem, we obtain an accuracy of 0.02°C over the typical body temperature range of 16-42°C. We report results from two preliminary trials regarding the impact of circadian rhythms and mental activity on skin temperature, indicating that our tool could indeed become a valuable asset for medical research.