Estimating core body temperature using electrocardiogram signals.

Suppressing the elevation in core body temperature is an important factor in preventing heatstroke. However, there is still no non-invasive method to sense core body temperature. This study proposed an algorithm that estimates core body temperature based on electrocardiogram signals. A total of 12 healthy men (mean age ± SD = 39.6 ± 13.4) performed an ergometric exercise load test under two conditions of exercise load in an environmental chamber adjusted to a temperature of 35°C and humidity of 50%. Vital sensing data such as electrocardiograms, core body temperatures, and body surface temperatures were continuously measured, and physical data such as body weight were obtained from participants pre- and post-experiment. According to basic physiological knowledge, heart rate and body temperature are closely related. We analyzed the relationship between core body temperature and several indexes obtained from electrocardiograms and found that the amount of change in core body temperature had a strong relationship with analyzed data from electrocardiograms. Based on these findings, we developed the amount of change in core body temperature estimation model using multiple regression analysis including the Poincaré plot index of the ECG R-R interval. The estimation model showed an average estimation error of -0.007°C (average error rate = -0.02%) and an error range of 0.457-0.445°C. It is suggested that continuous core body temperature change can be estimated using electrocardiogram signals regardless of individual characteristics such as age and physique. Based on this applicable estimation model, we plan to enhance estimation accuracy and further verify efficacy by considering clothing and environmental conditions.