Performance comparison of a wired zero-heat-flux sensor and wireless thermistor patch with esophageal probe for core temperature monitoring.

Accurate measurement of body temperature is essential in clinical practice; however, direct assessment of core temperature is invasive. Consequently, there is an increasing demand for non-invasive yet accurate methods. In this study, we compared two non-invasive techniques - a wired zero-heat-flux sensor (W⁺ZHF) and a wireless thermistor patch (W⁻TP) - against core temperature measurements obtained from an esophageal probe (EP). This prospective study included consecutive subjects who underwent esophagogastroduodenoscopy for health screening. The EP served as the reference for core temperature. Simultaneously, paired temperature readings were taken using W⁺ZHF and W⁻TP. Agreement between EP and each wearable device (W⁺ZHF and W⁻TP) was evaluated using subject-level Bland-Altman analysis, Pearson correlation coefficients, and intraclass correlation coefficients (ICC). Paired t-tests were used for descriptive comparison, and linear mixed-effects models were applied to account for repeated measurements within subjects. Among the 24 subjects, 5 (20.8%) were male and 19 (79.2%) were female, with a mean age of 32.8 ± 7.4 years. The mean temperatures measured by W⁺ZHF, W⁻TP, and EP were 36.56 ± 0.30°C, 36.61 ± 0.32°C, and 36.50 ± 0.26°C, respectively. Subject-level Bland-Altman analysis demonstrated a small bias between W⁺ZHF and EP (0.06°C; 95% confidence interval [CI], -0.04-0.15), with 95% limits of agreement (-0.42-0.53°C). The Pearson correlation coefficient between W⁺ZHF and EP was 0.63, and the ICC indicated moderate to high agreement (ICC, 0.62; 95% CI, 0.31-0.82). In contrast, W⁻TP showed greater variability in agreement with EP. Direct comparison between W⁺ZHF and W⁻TP revealed differences in agreement. These findings were consistent in linear mixed-effects models accounting for repeated measurements. Among the non-invasive devices evaluated, W⁺ZHF showed the closest agreement with EP measurements, although the predefined equivalence criterion was not formally met. These findings nonetheless suggest a promising role for zero-heat-flux technology in non-invasive estimation and support its further development in wearable devices.