Validity of the estimated core temperature algorithm during real-world prolonged walking exercise under warm ambient conditions

Abstract

Background

Non-invasive methods to estimate core body temperature (T_C) are increasingly available. We examined the group-level and individual participant-level validity of the Estimated Core Temperature (ECTemp™) algorithm to estimate T_C based on sequential heart rate (HR) measurements during real-world prolonged walking exercise in warm ambient conditions.

Methods

Participants walked 30 (n = 3), 40 (n = 13) or 50 (n = 2) km on a self-selected pace during which T_C was measured every minute using an ingestible temperature capsule. HR was measured every second and used to compute the estimated core temperature (T_C-est) using the ECTemp™ algorithm. Bland-Altman analyses were performed to assess agreement between T_C and T_C-est. A systematic bias <0.1 °C was considered acceptable.

Results

18 participants (56 ± 16 years, 11 males) walked for 549 min (range 418–645 min), while ambient temperature increased from 22 °C to 29 °C. Average HR was 108 ± 13 bpm and T_C ranged from 36.9 to 39.2 °C, whereas T_C-est ranged from 36.8 to 38.9 °C (n = 8572 observations). Group level data revealed a systematic bias of 0.09 °C (p < 0.001) with limits of agreements of ±0.44 °C. A weak correlation was found between T_C and T_C-est (r = 0.28; p < 0.001). Large inter-individual differences in bias (range −0.45 °C to 0.62 °C) and correlation coefficients (range −0.09 to 0.95) were found, while only 3 participants (17%) had an acceptable systemic bias of <0.1 °C.

Conclusion

Group level data showed that the ECTemp™ algorithm had an acceptable systematic bias during prolonged walking exercise in warm ambient conditions, but only 3 out of 18 participants had an acceptable systemic bias. Future studies are needed to improve the accuracy of the algorithm before individual users can rely on their estimated T_C during real-world exercise.