1 / 8 Case Study: Non-Invasive Core Body Temperature Measurements Case study: Non-invasive core body temperature measurements during sleep and daily life with greenteg gskin BodyTemp KIT. Lukas Durrer and Michele Zahner greenteg AG,, Contact: lukas.durrer@ 1. Abstract This case study presents body temperature measurements obtained with the gskin BodyTemp KIT. The measurements were taken on the forehead, on the chest and on the upper arm during sleep, daily life and physical activity. The gskin BodyTemp KIT is the world s first device that enables: Easy and accurate non-invasive and continuous core body temperature monitoring under free-living conditions Detection of important thermoregulatory events The gskin BodyTemp technology presented in this study can be easily integrated into any wearable device and will enable: Mobile sleep analysis Early diagnosis of health disorders affecting body temperature Heat stroke risk detection during sports and heavy work Ovulation tracking Investigation of thermoregulation and other physiological events Patient monitoring The current system works well on proximal measurement sites (chest, upper arm, forehead) and under varying environmental conditions. The measurement accuracy drops during physical activity and on distal measurement sites (wrist). greenteg is currently working on an improved version of the sensor hardware and algorithm that includes additional sensors, enabling accurate body temperature measurement at the wrist and during sports in the near future. 2. Introduction relevance of core body temperature measurements Early diagnosis of diseases is a powerful tool to increase quality of life and at the same time a promising way to stop or at least slow down the cost explosion in the health systems in developed countries. Wearable devices are seen as key technology for early diagnosis because they allow for continuously monitoring of key vital parameters of humans. Combining continuous monitoring of vital signs with state-of-the-art data analysis (big data or machine learning for example), early detection of health issues becomes possible. Besides heart rate, oxygen saturation, blood pressure and respiration rate, core body temperature (CBT) is one of the five most important vital parameters of a human being. While heart rate, oxygen saturation and respiration rate are commonly measured in the state-of-the-art wearable devices, non-invasive core body temperature and blood pressure sensors are not yet available. Several scientific studies have concluded that major health disorders such as chronic stress [Oka2001], insomnia [Lack 2008], Alzheimer s disease [Musiek 2015] and Parkinson s disease [Zhoung 2013] have repercussions on one s core body temperature. CBT is also an important indicator for healthy people when they travel (jetlag [Kojima 2013]), for athletes willing to maximize their performance during competition [Rosa 2016], for detecting ovulation [Baker2001] or for preventing heat stroke. According to scientific literature, a key parameter for non-invasive and continuous core body temperature measurement is the heat flux from the human body to the environment [Gunga 2008] and [Niedermann 2014].
2 / 8 Case Study: Non-Invasive Core Body Temperature Measurements greenteg produces highly sensitive and robust heat flux sensors that are small enough to be easily integrated into any wearable device and smart textile. For demonstration purposes, greenteg has developed a demonstration system, the gskin BodyTemp KIT, for measuring core body temperature at the forehead and the chest. 3. Measurement equipment The gskin BodyTemp KIT The gskin BodyTemp KIT consists of a standalone three-channel data logger, up to three gskin BodyTemp sensors and a readout software for downloading recorded measurement data via a USB linked to a PC (Figure 1). The BodyTemp sensors are provided with single-use dermal foam patches for a reliable attachment to the skin. This allows integrating the measurements into the daily routine of a person. Figure 1: gskin BodyTemp data logger and sensor
3 / 8 Case Study: Non-Invasive Core Body Temperature Measurements 4. Experimental greenteg has performed several measurement campaigns for validating the gskin BodyTemp KIT. This case study reports three measurements performed under different measurement conditions (see Table 1 for details): Test 1: Controlled environmental conditions (night in a sleep lab) Test 2: Free-living conditions Test 3: Physical activity The aim of these tests was to compare the core body temperatures calculated by the gskin BodyTemp KIT placed at three body locations (forehead, chest, upper arm) with an invasive reference measurement. An ingestible radio pill was selected for this purpose (VitalSense Capsule by Respironics Inc., USA, Equivital LifeMonitor, Hidalgo Ltd, UK) Measurement Procedure Participants ingested the radio pill between 6 to 8 h before starting the test to ensure the pill to be located in the intestine during the recording period. This measure minimizes temperature artifacts due to cold beverage or food intake. 15 minutes before starting the test the three available sensors were placed on the skin at their respective measurement position using a single-use foam patch provided by greenteg. Once the gskin BodyTemp KIT was set, the participant started the data recording. An initial resting period of ten minutes was needed for the autocalibration of the sensors to apply (forehead and chest). For distal measurement locations (i.e. upper arm), an inner body temperature value of 35 C was assumed as starting point. The gskin BodyTemp KIT continuously recorded every second. The Equivital LifeMonitor (Hidalgo Ltd, UK) was used to record the data of the radio pill temperature. The participant started the data collection simultaneously with the gskin BodyTemp KIT, taking a measurement every 15 seconds. Data analysis The agreement of the reference signal (radio pill) and the BodyTemp signals was quantified by an analysis of the Bland-Altman type. The mean of the difference signal (bias) and the standard deviation of the difference signal (SD) were both calculated. The limit of agreement was defined as ±1 SD.
4 / 8 Case Study: Non-Invasive Core Body Temperature Measurements Table 1. Test protocol and body locations selected for the gskin BodyTemp KIT in the three experiments reported (sleep, freeliving and activity). Test 1: Controlled environmental conditions The first measurement was recorded during the night, under controlled environmental conditions in a sleep lab. The gskin BodyTemp sensor was placed on the chest. The recording lasted 8 hours, starting from 11 PM. The participant spent the night in the sleep lab where he slept until 7 PM. Test 2: Free-living conditions In the second experiment, another participant was wearing a gskin BodyTemp sensor on the forehead and one on the middle of chest (sternum). The recording lasted 14 hours, starting from 5 PM. At the starting time the participant was still working in his office. The gskin BodyTemp KIT allowed the participant to finish his work (5 PM to 6.30 PM), to go outside and travel by train (6.30 PM to 7 PM), to arrive home and perform his daily domestic routine as usual before he went to bed at 11 PM and woke up at 7 PM. Test 3: Activity In the third experiment, another participant was wearing a gskin BodyTemp sensor on the forehead, middle of chest and upper arm. The recording consisted of 30 minutes of sitting on a desk followed by 15 minutes of continuous stepping (step bench) exercise. Finally, the participant cooled down for 30 more minutes at the desk.
5 / 8 Case Study: Non-Invasive Core Body Temperature Measurements 5. Results and Discussion Test 1: Controlled environmental conditions (sleep lab) Figure 2 shows the raw output signals of the gskin BodyTemp KIT (yellow line), the reference (radio pill, orange line) and the skin temperature (blue line) measured at the chest. The gskin BodyTemp sensors were positioned and the measurement started just before going to bed. The core body reading catches up with the radio pill reference only at 11 PM because the sensor takes 5 to 10 minutes to thermally equilibrate (on a covered chest sometimes up to 20min). After this equilibration period the non-invasively measured gskin BodyTemp signal follows the core pill signal with a bias of 0.16 C and limits of agreement of ± 0.25 C. Figure 2: Core body temperature measurements recorded with the gskin BodyTemp KIT at the chest under controlled environmental conditions. Orange: Radio pill measurement (reference); Yellow: non-invasive core body temperature measurement with the gskin BodyTemp KIT; Blue: skin temperature at the chest.
6 / 8 Case Study: Non-Invasive Core Body Temperature Measurements Test 2: Free-living conditions Figure 3 shows how the gskin BodyTemp KIT (green line) performed under real free-living conditions in comparison to the reference (radio pill, orange line) and the skin temperature (blue line) measured at the chest. The measurement started at 5 PM in the office. At 6.30 PM, the person left the office and went home, where he spent the evening with his family. At 11 PM, he went to bed. The non-invasive measurement has a bias of 0.13 C and limits of agreement of ±0.28 C. Figure 3: Core body temperature measurements during day and night under free-living conditions recorded with the gskin BodyTemp KIT placed at the chest and a radio pill. Overall, more than 10 overnight measurements on forehead and chest under free-living conditions were made. On average, the measurements had limits of agreement of ±0.25 C. This accuracy is perfectly adequate for clinical use [Sessler 2008 citation of figure 2].
7 / 8 Case Study: Non-Invasive Core Body Temperature Measurements Test 3: Compensation of environmental fluctuations The main advantage of using heat flux sensors for CBT determination is the ability to compensate environmental influences very efficiently (more than 80% compared to skin temperature measurements). This property can be easily seen in Figure Figure 3 (blue highlighted area). As the person went to bed and covered the sensor area (chest) with a blanket, the skin temperature started to steadily increase from 34.5 C to 36 C. However, the BodyTemp KIT was able to identify this skin temperature increase as resulting from a covering artifact and correctly predicted a slight decrease in body temperature, in agreement with the signal of the radio pill. The blue highlighted area in Figure 4 illustrates the behavior of the BodyTemp signal in case of faster and more pronounced changes in environmental conditions. In this particular measurement, the participant exited a well-heated building (23 C) and stepped outside (8 C), resulting in a rapid drop of the forehead skin temperature by 5.5 C (blue curve). greenteg s BodyTemp sensor (Error! Reference source not found.a orange curve) was able to almost fully compensate the rapid temperature drop and registered only a slight decrease of the (uncovered) head temperature. greenteg is working on an algorithm for exposed sensors that will tolerate rapid environmental changes even better. If the sensor is covered by clothing as the one on the chest shown in Figure 3 (same person), the current algorithm is already capable of compensating the less pronounced fluctuations in skin temperature. Figure 4: Same measurements as shown in Figure 3 measured at the forehead. Test 3: Activity We would like to use the third test to illustrate the limitations of the gskin BodyTemp KIT. As mentioned before, the algorithm is optimized for low activity conditions, like sleep or working on a desk, and if the sensors are positioned on the forehead and on the chest.
8 / 8 Case Study: Non-Invasive Core Body Temperature Measurements Figure 5 presents the gskin BodyTemp KIT readings at the forehead (red), chest (green) and upper arm (yellow) in comparison to the radio reference pill reading for a well-defined activity test protocol (see figure 5 below). 30 minutes after start time, all four curves highly correlated during working on the desk. As soon as the physical activity started, the core body temperature measured by the radio pill started to rise steadily, whereas the core body temperature obtained from the forehead and the upper arm showed a slight initial decrease (probably convective cooling due to the body movement). However, the most prominent artifact is the high overshoot of all three gskin BodyTemp temperature signals to 41 C and 42 C, while the core body temperature maximum observed by the radio pill did not exceed 38.5 C. Figure 5. Activity test: Thermophysiological measurement with the gskin BodyTemp KIT. Body temperature recorded during stepping activity (step bench) at the forehead (red), chest (green) and upper arm (yellow) in comparison to the reference ingestible thermometer enclosed in the radio pill (blue). The thermoregulatory reactions of the body during physical activity are very complex. In order to detect and compensate for all thermoregulatory events during activity, additional physiological parameters have to be integrated into the BodyTemp system. These parameters can be measured using additional non-invasive sensors available commercially. These sensors signals are not yet part of greentegs gskin BodyTemp KIT. 6. Conclusions The gskin BodyTemp KIT from greenteg is able to monitor core body temperature accurately, non-invasively and continuously during sleep and the daily routine if sensors are placed on the forehead and on the chest. Although promising at this stage, the use of the gskin BodyTemp KIT for thermoregulatory research i.e. circadian rhythm and REM sleep phase detection requires further verification. Accurate core body temperature detection during increased physical activity and at distal body positions (upper arm and wrist) requires a more elaborated algorithm in combination with additional sensors.