Circadian Phase Assessment of Core Body Temperature Using a Wearable Temperature Sensor Under the Real World

Abstract

Objective

To evaluate whether a patch-type wearable temperature sensor (CALERA Research) could determine the circadian phase of core body temperature (CBT) in a manner like a rectal probe.

Objective

Sixteen participants (27 ± 11 years, 8 males and 8 females) wore an actigraph and CALERA Research sensor on the chest region for 3–5 days in a real-world setting. Simultaneous rectal temperature measurements were performed during the nocturnal sleep period. The midpoints of the nocturnal decrease in CBT (CBT_trough) were used as the circadian phase marker. We analyzed 60 pairs of CBT_trough. The reliability and agreement of the CBT_trough from the two devices were analyzed using the intraclass correlation coefficient (ICC) and the concordance correlation coefficient (CCC). The Bland-Altman analysis was used to quantify the limit of agreement of CBT_trough between the devices.

Objective

The ICC of 0.96 (95%CI: 0.93–0.98) and CCC of 0.96 (95%CI: 0.93–0.97) values indicated excellent reliability and substantial agreement, respectively. The mean bias was 0.16 hours (95%LoA: -0.76–1.07 hours). The mean CBT_trough comparison was 5.9 ± 1.6 hours in the CALERA Research sensor and 5.8 ± 1.7 hours in the rectal probe.

Conclusion

The difference in the CBT_trough between the two devices was about ± 1.0 hour which would be an acceptable range for determining the CBT_trough. We suggest that the CALERA Research sensor could be a useful tool for reasonably estimating the circadian phase of CBT_trough and providing a surrogate for a rectal probe.

 

MissingLinkText

Keywords

circadian rhythm - sleep - rectal temperature - patch-type sensor

Introduction

The circadian rhythm of the core body temperature (CBT) is closely associated with diurnal changes in physiological and cognitive functions. This CBT rhythm is regulated by a central circadian pacemaker located in the suprachiasmatic nucleus (SCN).[1] Assessing the circadian phase of the core body temperature (CBT) is important for adjusting an individual's circadian rhythm by exposure to a timed bright light, so-called bright-light therapy. The human circadian system comprises two distinct circadian oscillators that separately regulate the circadian rhythm of rectal temperature and the sleep-wake cycle.[2] [3] Thus, the phase difference between the circadian phase of the CBT, as well as the sleep-wake cycle of these two oscillators, provides valuable information. Patients with circadian rhythm sleep disorders, such as non-24-hour sleep-wake disorder and delayed sleep-wake disorder, have altered phase relationships between the circadian rhythms of their CBT and the sleep-wake cycle.[4] Moreover, misalignment between the circadian rhythm and sleep-wake cycle increases the incidence of various diseases.[5] Thus, assessing the circadian phase of CBT could allow us to evaluate the internal phase relationship between the circadian pacemaker in the SCN and the sleep-wake cycle.

Rectal temperature is commonly used as a marker to assess the circadian phase of CBT in the research and clinical fields of sleep and chronobiology.[3] [6] [7] The rectal temperature is often measured once per minute by using a flexible wired rectal probe. Recently, some noninvasive temperature sensors, such as ingestible capsule sensors[8] and ear canal sensors[9] have been used for monitoring CBT. In addition, a patch-type wearable temperature sensor has been developed and used for monitoring the CBT during sleep periods,[10] exercise conditions,[11] and when in a hot environment.[12] [13] [14] Regarding the accuracy of the wearable temperature sensor, some studies reported relatively low accuracy for CBT values.[11] [12] [13] [14] In human sleep and circadian rhythm research, patch-type wearable temperature sensors would have been expected to easily assess the circadian phase of the CBT and provide a surrogate for a rectal probe. However, there is still a lack of research focused on assessing the circadian phase of the CBT using the patch-type wearable temperature sensor in participants in the real world. In the present study, we performed comparative measurements of CBT using the patch-type wearable temperature sensor and a rectal probe from healthy adults in the real world. Here, we demonstrated that a wearable temperature sensor could be a useful tool for reasonably estimating the circadian phase of the rhythm sleep disorders, such as CBT_trough and provide a surrogate for a rectal probe.

 

MissingLinkText

Materials and Methods

Participants

A total of 16 (8 males and 8 females) participants aged 19–45 years participated in the present study as paid volunteers. None of the participants worked early in the morning, late at night, or on rotating night shifts. None of the participants had any history of psychiatric, endocrine, or sleep disorders. All participants provided written informed consent before participating in the study and were able to withdraw from the experiment at any time. This study was approved by the Ethics Committee of the Hokkaido University Graduate School of Education (#19–10) and conducted by the Declaration of Helsinki.

 

MissingLinkText

Experimental Protocols

To estimate sleep and wakefulness, all participants were asked to use an actigraph worn on the wrist. The participants were also asked to complete a sleep diary. In addition, the participants were instructed to wear the CALERA Research sensor outside of bathing to measure CBT. Moreover, rectal temperature was measured intermittently from 3–4 hours before bedtime to 1–3 hours after waking up each day. The participants were prohibited from exercising and taking showers and/or baths during rectal temperature measurements. All data were measured for the participants under free-living activities for 3–5 days. [Fig. 1A] illustrates location of the sensors on a participant's body and the example of experimental protocol ([Fig. 1B]).

Fig. 1 Location of the sensors on a participant's body and example of experimental schedule. The left figure (A) indicates the location of the sensors on a participant's body. An actigraphy device (ActTrust2) was worn on the non-dominant wrist. The wearable temperature sensor (CALERA Research sensor) was attached to the torso at ∼20 cm below the armpit, and the rectal probe was inserted into the rectum from the anal sphincter. Right figure (B) shows an example of an experimental schedule for 3 days. Each arrow shows the recording periods of each sensor.

 

MissingLinkText

Measurements

Sleep-wake Cycle

Wrist activity was recorded every minute using an actigraph worn on the nondominant wrist (ActTrust2; Condor Instruments, São Paulo, Brazil). We used the activity count measured by the wrist actigraph to estimate the participants' sleep and waking states. The activity count was automatically scored using the built-in Cole–Kripke algorithm[15] in Act Studio software ver. 1.0.24 (Condor Instruments).

 

MissingLinkText

Patch-type Wearable Temperature Sensor

A patch-type, wearable, and commercially available temperature sensor, CALERA Research sensor (greenTEG), was used in the present study to estimate the circadian phase of the CBT. The CALERA Research sensor assesses skin temperature and heat flux and employs machine learning technology with an AI algorithm to estimate the CBT. The CBT data estimated using the CALERA Research sensor were automatically transmitted and stored on a web server (CORE cloud). Original body temperature data (1-minute bins) were downloaded to a computer as .csv files and were used to analyze the circadian phases of the CBT. The participants were instructed to wear a CALERA Research sensor attached to the torso at ∼20 cm below the armpit by the manufacturer's recommendations. The CALERA Research sensor was worn on the skin using a medical-grade adhesive patch (A-165210, green TEG).

 

MissingLinkText

Rectal Temperature Probe

A rectal probe (401J; Yellow Springs Instrument Co., Inc. USA) was used for measurements. Rectal temperature data (1-minute bins) were measured and stored using a data logger (NT-logger N543R; Nikkiso Thermo). The participants were instructed to insert the rectal probe 15 cm beyond the anal sphincter by themselves.

 

MissingLinkText

 

MissingLinkText

Data Analysis

The trough phase of the CBT was defined as the midpoint of nocturnal decrease in CBT (CBT_trough) and used as the circadian phase markers, as already established.[16] [17] The CBT_trough was determined by a geometric method as previously published[16] [17] with slight modification, as below. Briefly, the original temperature data for every 1-minute bin were averaged at 10-minute intervals and smoothed using a three-point moving average method. The CBT_trough was then defined as the middle of two points where a line on the middle level between the temperature at the point more than 0.2°C higher from the minimum values of body temperature crossed the descending and ascending parts of the temperature rhythm ([Fig. 2]). In the above analysis, two of the 16 participants were excluded because of missing temperature data (e.g., the rectal probe slipped out and/or a nocturnal drop in body temperature was not observed). Therefore, temperature data from 14 participants were used in the analysis. We successfully collected 34,537 data points (1-minute bins) and 60 circadian phases of the CBT_trough measured using the CALERA Research sensor and rectal probe. To evaluate actual temperature values between the two devices, body temperatures during the nocturnal sleep period were averaged and compared between the CALERA Research sensor and rectal probe.

Fig. 2 Geographic method to determine the circadian phase of CBT. The left panels indicate representative recordings of CBT measured by the CALERA Research sensor (grayed line) and rectal probe (solid line) before, during, and after the nocturnal sleep period in the participant (male, 22 years). The geographic method used to determine the CBT_trough phase (dashed lines and arrows) is illustrated. The gray areas show the actigraphy-based sleep periods. CBT: core body temperature.

 

MissingLinkText

Statistical Analysis

The reliability of the CBT_trough measured by the CALERA Research sensor was analyzed by the intraclass correlation coefficient (ICC) of the CBT_trough in a rectal probe. The ICC is a value between 0 and 1, where a value less than 0.5, between 0.5 and 0.75, between 0.75 and 0.9, and greater than 0.90 indicate poor, moderate, good, and excellent reliability.[18] Lin's concordance correlation coefficient (CCC)[19] was used to evaluate the agreement and reliability between the CBT_trough measured by the CALERA Research sensor and rectal probe. A CCC value less than 0.90, between 0.90 and 0.95, between 0.95 and 0.99, and greater than 0.99 indicates poor, moderate, substantial, and almost perfect agreement, respectively.[20] Furthermore, the Bland-Altman plot[21] was used to compare the CBT_trough between the CALERA Research sensor and rectal probe and calculate the mean difference and 95% limits of agreement (95% LoA). Statistical analyses were performed using StatView software (version 5.0, SAS Institute, Cary, NC, USA) and R software (version 4.4.0, The R Foundation for Statistical Computing, Vienna, Austria).

 

MissingLinkText

 

MissingLinkText

Results

The participants' characteristics and the number of circadian phases used in the present analysis are shown in [Table 1]. [Fig. 3] shows representative recordings of wrist activity and body temperature measured by the CALERA Research sensor and rectal probe. The ICC value of 0.96 (95%CI: 0.93 to 0.98) value indicated excellent agreement,[18] whereas the CCC value of 0.96 (95%CI: 0.93 to 0.97) values indicated substantial agreement[20] between the two devices. The Bland-Altman analysis found the mean bias of CBT_trough between the two devices was 0.16 hour (95%LoA: -0.76 hour to 1.07 hour) ([Fig. 4]). The mean of the CBT_trough was 5.9 ± 1.6 hour in CALERA Research sensor and 5.8 ± 1.7 hour in rectal probe, respectively. During the nocturnal sleep period, the mean body temperature measured by the CALERA Research sensor was 36.69 ± 0.16°C, while the rectal probe measured 36.59 ± 0.25°C, indicating that the CALERA Research sensor overestimated the body temperature by ∼0.10°C (95% CI: 0.06 to 0.14°C) as compared with a rectal probe. The differences in the mean temperature values between the two devices were not consistent in the 60 records. The mean temperature values in the CALERA Research sensor were higher than those in the rectal probe (46 of 60 records, 77%). In contrast, the other 14 records (23%) showed lower temperature values in the CALERA Research sensor as compared with the rectal probe. In addition, the difference in the temperature value between the two devices showed inter- and intra-individual differences. Six of 14 participants (43%) showed higher temperature values in the CALERA Research sensor as compared with the rectal probe, and the other 8 participants (57%) showed higher or lower temperature values in the CALERA Research as compared with the rectal probe.

Fig. 3 Representative recording of CBT measured by the CALERA Research sensor and rectal probe with simultaneous recording of wrist activity in the real world for 3 days. The top panel indicates a representative recording of CBT measured with the CALERA Research sensor (grayed line) and rectal probe (solid line). The lower panel indicates wrist activity measured using wrist actigraphy. The gray areas show the actigraphy-based sleep periods.

Fig. 4 Bland—Altman plot comparison for the CBT_trough phase determined by the CALERA Research sensor and rectal probe. The solid line indicates the mean difference between the CBT_trough phase determined by the CALERA Research sensor and rectal probe. The dotted lines indicate 95% confidence intervals.

Table 1 Participants' demographic characteristics, recording period, and number of CBT phases

Participants	Age (yrs.)	Sex	BMI (kg/m2)	Recording period (dd/mm/yy)			Number of CBT phases	Sleep onset (h, local time)	Sleep offset (h, local time)
1	42	M	24.0	11/8/22	−	28/8/22	5	0.1 ± 2.7	6.9 ± 0.5
2	45	F	19.3	28/10/22	−	13/11/22	4	2.8 ± 0.4	8.5 ± 0.9
3	43	F	24.0	7/12/22	−	14/12/22	5	2.0 ± 1.0	7.5 ± 0.7
4	19	F	19.7	13/12/22	−	20/12/22	4	2.0 ± 0.6	7.8 ± 0.4
5	22	M	19.7	12/12/22	−	17/12/22	4	1.2 ± 0.2	8.6 ± 0.4
6	48	M	19.4	3/1/23	−	8/1/23	5	23.7 ± 0.3	6.5 ± 0.8
7	22	M	18.3	6/1/23	−	10/1/23	3	2.4 ± 1.4	9.2 ± 2.4
8	22	F	17.2	9/1/23	−	16/1/23	3	2.4 ± 1.2	11.4 ± 0.4
9	21	F	19.1	27/12/22	−	2/1/23	5	0.4 ± 1.1	8.1 ± 0.5
10	19	F	22.7	28/12/22	−	2/1/23	N/A
11	21	F	24.5	10/1/23	−	15/1/23	5	2.9 ± 1.1	9.4 ± 1.2
12	25	F	23.1	10/1/23	−	21/1/23	N/A
13	23	M	21.1	8/7/23	−	3/9/23	4	1.0 ± 1.0	8.3 ± 1.7
14	21	M	17.7	28/6/23	−	29/7/23	4	2.3 ± 1.0	9.1 ± 0.8
15	21	M	20.6	9/7/23	−	5/9/23	4	0.6 ± 0.2	8.7 ± 1.0
16	20	M	22.4	9/7/23	−	13/7/23	5	2.2 ± 0.3	9.9 ± 0.9
Mean	27.1		20.8					1.0	8.5
SD	10.5		2.4					1.4	1.4

Abbreviations: BMI, body mass index; CBT, core body temperature; N/A, not applicable.

Data from 2 (#10 and #12) of 16 participants who did not assess the phases of core body temperature were excluded from the analysis.

 

MissingLinkText

Discussion

The present study aimed to evaluate whether a patch-type wearable temperature sensor could be used to assess the circadian phase of CBT, like the use of a wired rectal probe from healthy adult participants in the real world. Although the absolute temperature value during the sleep period was different between the two devices, body temperature at the same time points was similar enough to use for circadian determination. Thus, the geometric method could be applied to determine the circadian phase of the CBT measured by both devices. The ICC and CCC values indicated the agreement of the CBT_trough estimated by the CALERA Research sensor and rectal probe was excellent and substantial agreement, respectively. The result of the Bland-Altman analysis demonstrated that the 95% LoA of the mean difference ranged between the CBT_trough of CALERA Research sensor and rectal probe was -0.76 hour to 1.07 hour, indicating that there was about ± 1 hour difference in the CBT_trough between the CALERA Research sensor and rectal probe. The mean temperature values during the sleep period were higher in the CALERA Research sensor by 0.1°C on average as compared with the rectal probe.

In humans, the candidates for physiological functions (melatonin, cortisol, and CBT) reflect the phase of the central circadian pacemaker in the SCN.[22] Assessing the circadian phase of SCN pacemakers in circadian rhythm and sleep research in clinical settings usually relies on circadian rhythms of melatonin and/or CBT. The most reliable marker of the circadian pacemaker in humans is melatonin levels in blood (plasma or serum) or saliva. Although the circadian rhythm of melatonin is the most reliable circadian phase marker for assessing the circadian rhythms in humans, sample collection (every 30–60 minutes) from subjects under dim light conditions is required.[23] [24] In contrast to measuring the circadian rhythm of melatonin, measuring CBT remains useful for assessing the phase of circadian rhythms in humans.[22] However, the measurement of the CBT by using a wired rectal probe is sometimes difficult for some subjects such as children and unconscious patients under the real world or clinical settings. In contrast to a wired rectal probe, a wireless wearable temperature sensor patched on the skin surface could allow us to assess the CBT_trough regardless of the participant's age and disease. Under clinical and practical conditions, bright-light therapy is an effective treatment for seasonal affective disorders and circadian rhythm-related sleep disorders. Appropriate timings for bright-light exposure are commonly determined using the circadian phase of the CBT and the phase response curve to a single bright-light.[25] [26] The reported phase response curve to a single bright-light has an advanced (delay) portion of 6 hours after (before) showing the CBT_trough. In the present analysis, the CBT_trough estimated by the CALERA Research sensor was about ± 1 hour different from the CBT_trough measured by the rectal probe. Therefore, the CALERA Research sensor would be useful to provide the circadian phase marker to determine the optimum time of day of light exposure when conducting not only bright-light therapy but also exogenous melatonin ingestion.

Of note, it has been demonstrated that circadian rhythms in CBT are attenuated or occasionally disappear in subjects showing internal desynchronization between the circadian pacemaker in the SCN and the sleep-wake cycle in the real world.[27] Therefore, it has been recommended that the circadian rhythm of plasma melatonin or cortisol levels, rather than the CBT, is used to assess the circadian phase more precisely in participants undergoing internal desynchronization.[22] [27]

The present study had some methodological limitations. First, it should be noted that the temperature value measured by the CALERA Research sensor still requires validation, and the accuracy and reliability need to be improved.[11] [12] [13] [14] Second, the means of CBT values during the sleep period was higher in the CALERA Research sensor by 0.10°C as compared with the CBT in the rectal probe. This discrepancy in the CBT values between the rectal temperature and the CALERA Research sensor was not consistent and showed inter- and intra-individual differences. It is necessary to examine various environmental and physiological factors to influence the temperature value in the CALERA Research sensor. Lastly, the data was obtained from a relatively small number of participants. Further studies should confirm our results in a larger group of healthy participants and patients such as those with the circadian rhythm sleep or seasonal affective disorders.

 

MissingLinkText

Conclusion

This study demonstrates that a patch-type, heat-flux-based wearable temperature sensor could be an alternative method for assessing the circadian phase of CBT in healthy adult participants in the real world.

 

MissingLinkText

 

MissingLinkText

Conflict of Interests

The authors have no conflict of interest to declare.

Acknowledgments

The authors are grateful for the valuable contributions of all study participants.

• References

• 1 Saper CB, Lu J, Chou TC, Gooley J. The hypothalamic integrator for circadian rhythms. Trends Neurosci 2005; 28 (03) 152-157
• 2 Honma K, Hashimoto S, Endo T, Honma S. Internal desynchronization in the human circadian rhythm. Sapporo, Japan: Hokkaido University Press; 1998: 101-113
• 3 Wever RA. Influence of physical workload on freerunning circadian rhythms of man. Pflugers Arch 1979; 381 (02) 119-126
• 4 Uchiyama M, Okawa M, Shibui K. et al. Altered phase relation between sleep timing and core body temperature rhythm in delayed sleep phase syndrome and non-24-hour sleep-wake syndrome in humans. Neurosci Lett 2000; 294 (02) 101-104
• 5 Baron KG, Reid KJ. Circadian misalignment and health. Int Rev Psychiatry 2014; 26 (02) 139-154
• 6 Aschoff J. Circadian rhythms in man. Science 1965; 148 (3676) 1427-1432
• 7 Czeisler CA, Duffy JF, Shanahan TL. et al. Stability, precision, and near-24-hour period of the human circadian pacemaker. Science 1999; 284 (5423) 2177-2181
• 8 Micic G, Lovato N, Gradisar M, Burgess HJ, Ferguson SA, Lack L. Circadian melatonin and temperature taus in delayed sleep-wake phase disorder and non-24-hour sleep-wake rhythm disorder patients: An ultradian constant routine study. J Biol Rhythms 2016; 31 (04) 387-405
• 9 Choi Y, Nakamura Y, Akazawa N. et al. Effects of nocturnal light exposure on circadian rhythm and energy metabolism in healthy adults: A randomized crossover trial. Chronobiol Int 2022; 39 (04) 602-612
• 10 Xu X, Wu G, Lian Z, Xu H. Feasibility analysis of applying non-invasive core body temperature measurement in sleep research. Energy Build 2024; 303: 113827
• 11 Daanen HAM, Kohlen V, Teunissen LPJ. Heat flux systems for body core temperature assessment during exercise. J Therm Biol 2023; 112: 103480
• 12 Goods PSR, Maloney P, Miller J. et al. Concurrent validity of the CORE wearable sensor with BodyCap temperature pill to assess core body temperature during an elite women's field hockey heat training camp. Eur J Sport Sci 2023; 23 (08) 1509-1517
• 13 Jolicoeur Desroches A, Naulleau C, Deshayes TA, Pancrate T, Goulet EDB. CORE™ wearable sensor: Comparison against gastrointestinal temperature during cold water ingestion and a 5 km running time-trial. J Therm Biol 2023; 115: 103622
• 14 Verdel N, Podlogar T, Ciuha U, Holmberg HC, Debevec T, Supej M. Reliability and validity of the core sensor to assess CORE body temperature during cycling exercise. Sensors (Basel) 2021; 21 (17) 5932
• 15 Cole RJ, Kripke DF, Gruen W, Mullaney DJ, Gillin JC. Automatic sleep/wake identification from wrist activity. Sleep 1992; 15 (05) 461-469
• 16 Endo T, Honma S, Hashimoto S, Honma K. After-effect of entrainment on the period of human circadian system. Jpn J Physiol 1999; 49 (05) 425-430
• 17 Hashimoto S, Nakamura K, Honma S, Tokura H, Honma K. Melatonin rhythm is not shifted by lights that suppress nocturnal melatonin in humans under entrainment. Am J Physiol 1996; 270 (5 Pt 2): R1073-R1077
• 18 Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med 2016; 15 (02) 155-163
• 19 Lin LI. A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989; 45 (01) 255-268
• 20 GB M.. A proposal for strength-of-agreement criteria for Lin's concordance correlation coefficient. NIWA Client Report: HAM2005–062. 2005
• 21 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1 (8476) 307-310
• 22 Honma K, Yamanaka Y. Circadian rhythm measurements in humans. Hirota T, Hatori M, Panda S. , editor. New York: Neuromethods 2022; 186: 1-27
• 23 Benloucif S, Burgess HJ, Klerman EB. et al. Measuring melatonin in humans. J Clin Sleep Med 2008; 4 (01) 66-69
• 24 Pandi-Perumal SR, Smits M, Spence W. et al. Dim light melatonin onset (DLMO): a tool for the analysis of circadian phase in human sleep and chronobiological disorders. Prog Neuropsychopharmacol Biol Psychiatry 2007; 31 (01) 1-11
• 25 Honma K, Honma S. A human phase response curve for bright light pulse. Jpn J Psychiatry Neurol 1988; 42: 167-168
• 26 Minors DS, Waterhouse JM, Wirz-Justice A. A human phase-response curve to light. Neurosci Lett 1991; 133 (01) 36-40
• 27 Nakamura K, Hashimoto S, Honma S, Honma K, Tagawa Y. A sighted man with non-24-hour sleep-wake syndrome shows damped plasma melatonin rhythm. Psychiatry Clin Neurosci 1997; 51 (03) 115-119

Publication History

Received: 29 August 2024

Accepted: 11 November 2024

Article published online:
30 January 2025