Abstract
The development of alternative methods for monitoring cardiorespiratory function without restraint or surgical implantation is attracting growing interest for both ethical and scientific reasons. For this purpose, a new non-invasive jacketed telemetry tool consisting in a radio device maintained in a jacket worn by the animal was previously developed to improve cardiorespiratory monitoring. It allows simultaneous monitoring of cardiac activity by surface electrocardiagram, respiratory function by respiratory inductive plethysmography, and locomotor activity by accelerometry. However, this tool has only been validated under conditions of low/intermediate activity levels or in anesthetized animals. This study aimed to evaluate the feasibility of using this system in the challenging conditions of an exertion protocol. Male Wistar rats (n = 10, 8–9 weeks old) were subjected to an incremental treadmill exercise protocol including speed levels from 5 to 40 cm s−1 separated by 30-s breaks. Heart rate (HR) and minute ventilation (assessed by minute volume; MV) were continuously monitored. At the end of each running level and during the 30-s breaks, HR and MV showed a significant increase compared to resting values. They returned to the baseline within 60 min of post-exercise recovery. Overall, our results demonstrated (i) the ability of the animal to run while wearing the device and (ii) the ability of the device to reliably monitor cardiorespiratory adaptation to treadmill exercise despite significant mechanical disturbances. In conclusion, this study highlights the possibility of non-invasively monitoring cardiorespiratory functional variables that were previously unattainable under conditions of high activity in freely moving animals.
Introduction
Monitoring cardiac and ventilatory functions is a crucial aspect in preclinical and clinical environments. However, acquiring relevant physiological data often remains a technological challenge. In particular, to ensure the physiological value of the data, their acquisition must be as stealthy as possible, so as not to disturb the function being explored. Therefore, cardiac and ventilatory functions should ideally be monitored in real life conditions, without interfering with the activities of the patient or animal under study, and without being affected by motion artifacts. In the field of preclinical research, there is currently no measuring device or instrument that fully meets the specifications described previously, and in particular that complies with the ethical rules of the 3Rs. Rodent species are widely used in regulatory studies, as well as in pharmacology. A high-performance device adapted to monitoring ventilation in these species, especially under conditions of high-level motion (e.g., exercise protocols) would therefore be of great interest.
Today, most methods available for directly assessing cardiac and ventilatory functions in rodents cannot be used effectively in exercise protocols due to practical limitations. For example, unrestrained whole body plethysmography chambers, which are the gold standard for complete access to ventilatory function in awake animals, 1 can only be used in stationary animals. The use of face masks remains limited to anesthetized animals only, although some trials at exercise were reported. 2 Similarly, implanted radiotelemetry, the gold standard for the study of cardiac function in free-moving rodents3,4 can be used to monitor electrocardiogram (ECG) and haemodynamic parameters5 –7 but comes with important constraints essentially linked to the surgical implantation of sensors and transmitter. In particular, postoperative pain and impeded movement due to the surgical wounds or to the implanted device itself are critical in conditions of movement and have to be managed with appropriate care protocols. 8 In this context of a growing interest in improving the monitoring of cardiorespiratory parameters in free-moving animals, an increasing number of studies are focusing on non-invasive methods. Studies have been conducted with wearable ECG radiotelemetry devices, 9 a wristwatch-like actimetry device, 10 a Doppler radar-based respiratory activity detecting device, 11 and wearable biosensors, 12 but most of these monitoring techniques have so far been limited to the proof-of-concept stage, and need to be further developed before they can be used on a larger scale. Recent research has also demonstrated the ability of systems based on RGB video cameras to explore the respiratory activity of completely unconstrained mice 13 and rats. 14 The main restrictions of these technologies are that they are limited to inactivity phases of the animals and only provide access to respiratory rate. Jacketed telemetry, which is commonly used as a non-invasive solution for cardiorespiratory assessment in large animals (dogs, primates, etc.)15,16 has been recently adapted to laboratory rats17,18 with the development of the DECRO system (ETISENSE, Lyon, France). This system consists of a non-invasive and low-constraint radio device maintained in a jacket worn by the animal. It allows combined monitoring of cardiac activity by surface ECG, respiratory function by respiratory inductive plethysmography (RIP), and locomotor activity by accelerometry. Previous studies have demonstrated the ability of this system to effectively monitor cardiorespiratory variables against gold standards17 –19 in a variety of conditions, such as socialization in living conditions or in the context of pharmacological studies.20,21 So far, the situations under which the device has been validated remain limited to low levels of activity resulting in moderate noise conditions. The ability of DECRO to provide relevant data under intense motion conditions remains to be demonstrated, which would considerably broaden its applicability in preclinical research.
Therefore, the present study aimed to evaluate the feasibility of using DECRO in the challenging condition of an exertion protocol. To this end, a progressively increasing intensity treadmill exercise was imposed on Wistar rats to evaluate both the ability of the animals to run on the treadmill while wearing the telemetry system, and the capacity of the device to accurately capture the adaptive cardiorespiratory response under moving conditions.
Materials and methods
Animals
Male Wistar rats (n = 10, 8–9 weeks old, 270–300 g, from Charles River, Lyon, France) were housed in pairs in conventional cages (Techniplast, Louvier, France) and placed in a standard laboratory animal facility (pressure = 1 ± 0.5 bar, T° = 20°C to 24°C, hygrometry = 40% to 70%, light/dark cycle = 12 h/12 h) with free access to water and food. The litter was replaced twice a week. Enrichment included nesting material, gnawing objects and tunnels (Comptoir des sciures, Meyzieu, France).
The animals were weighed and marked for identification on arrival (ink marks on the tail). The clinical status of the animals was monitored throughout the study using behavioral criteria (fur condition, posture, isolation) and daily weight assessment.
Ethical concerns
All experiments were conducted in compliance with the European (L276-33 2010/63/EU) and French (AGRG1231951D) regulations concerning the protection and use of laboratory animals and complied with the recommendations the Declaration of Helsinki.
This study was approved by the ethics committee #012 (Grenoble Alpes University) accredited by the French Ministry of Higher Education and Research (referenced APAFIS#31347-2021042914221915 v2).
Non-invasive telemetric ECG and respiratory inductive plethysmography recording
Cardiac, respiratory, and locomotor activity variables were measured using a Bluetooth-telemetric jacket system (DECRO, ETISENSE, Lyon, France). Respiratory function was monitored using RIP sensors integrated directly into the jacket to measure instantaneous changes in trunk volume. ECG electrodes were placed under the jacket to monitor cardiac electrical activity and a tri-axis accelerometer integrated in the electronic device worn in the back of the animal provided a direct measurement of locomotor activity.
A 15 g Bluetooth transmitter (emitter + battery) was secured in a backpack integrated into the jacket to send data to a receiver located in the same room. This receiver included LASA software (v2.0.0, ETISENSE, Lyon France) for the acquisition and analysis of data.
Experimental protocol
Habituation
After 1 week of acclimatization, the animals underwent the phase of habituation to the device as described by Fares et al. 22 and implemented in previous studies with the device.20,21 On the first day, the animals were shorn on the back before the first fitting to allow the placement of two flexible dry gel electrodes (Ag/AgCl) for ECG recording. Afterwards, they were accustomed to wear the jacket and the electronic device for 3 days. The fur clipping was repeated once a week for the duration of the protocol. A second phase, involving a gradual increase in running speed and exercise duration, was carried out over a 2-week period, 5 days a week. Controlled physical exercise was applied to the animals using a single-track motorized treadmill (BIOSEB, France) delivering electric shocks of (0.2 mA) to motivate the animals to run during the early stages of the protocol. If the limit of 10 shocks in one session was reached, the session was terminated. Once the animal was equipped with the system, it was first placed in its cage (20 min) and then on the stopped treadmill (3 min). After a first slow speed stage (5 m scm s−1, 5 min), the animals were submitted to progressive increases in running speed of 5 m scm s−1 every 2 min for a duration of 5 (on the first day of habituation) to 25 min (last day of habituation). At the end of the exercise, the animal was returned to its cage, and was de-equipped after 30 min. The running speed reached as well as the number of shocks were monitored to adjust the intensity and duration of the exercise over the 10-day habituation.
Exercise protocol
An incremental exercise protocol derived from that described by Qin et al. 23 was chosen, based on a gradual increase in the speed of the treadmill. The initial protocol was customized with the addition of 30-s immobility phases (treadmill stop) before each speed increase. These short breaks were intended to enable the comparison of the variables captured between the two situations of movement and immobility. The exercise protocol began with a post-equipment stabilization in the cage (1 h) before placing the animal on the treadmill for 3 min at rest before baseline (BL) measurements. The animal was then subjected to five speed levels at 5, 10, 20, 30, and 40 cm s−1 (5 min/level) each separated by a 30-s break. Then, the rats were returned to their cages and cardiac and respiratory functions were monitored for 1 h of recovery.
Data recording and analysis
Physiological variable recording
The recording started once the animal had been equipped with the jacket and ended after 1 h of post-exercise recovery. Heart rate (HR), expressed in beats per minute (bpm) was calculated from the ECG recording. Ventilatory function was assessed by minute volume (MV) in milliliters per minute per kilogram body weight (ml min−1 kg−1) and respiratory rate (RespR) in cycles per minute (cpm). The locomotor activity level (AL) in milli-gn (mgn, with 1 gn corresponding to 9.81 m s−2) defined
Data processing
The software was used to automatically detect ventilatory and cardiac cycles (respectively delimited by the end of expiration (Vmin) for the RIP signal and the R-peaks for the ECG signal) and to calculate the corresponding physiological variables. Non-valid cycles were automatically excluded by the detection module of the software according to the default exclusion ranges that have to be configured by the experimenter. In this work, we defined broad thresholds for the detection of the RR interval (from 85 ms to 6 s), tidal volume (from 0.1 to 4 arbitrary units), and respiratory cycle duration (from 0.15 to 10 s). The variables were then averaged every 5 s and reported as mean ± SEM. For each animal, signal quality was reviewed and the tagged periods of interest were marked in the LASA software. BL was defined as the 30-s rest period just before the animal was placed on the treadmill (end of post-equipment stabilization and locomotor AL below 10 mgn). During the exercise protocol, data were analyzed during immobility (“break”) and running phases (“run”) in order to compare the values in the two situations and evaluate the potential effect of movement artifacts. Break and run analysis timeframes for every speed were located using locomotor AL signals and marked in the software. Break timeframes were defined as periods of 30-s following the immediate decrease of the locomotor AL signal induced by the stop of the treadmill. The run timeframes were defined as the final 30 s of each speed level. In the case of an animal that failed to finish the run speed level, data were excluded from this speed level and for the following ones. Data during post-exercise recovery were averaged over 30-s intervals every 10 min between 0 and 60 min.
Finally, for each ECG and RIP-derived variables the software provides a detection score defined as the percentage of cumulated duration of valid ventilatory or cardiac cycles over the recording period. This detection score reflects the exclusion of invalid ventilatory or cardiac cycles by the software’s algorithms based on physiological criteria (minimum and maximum tolerated tidal volume, respiratory or cardiac rate, airflow, etc.). Any cycle outside of these tolerance ranges was rejected resulting in a reduced detection score. This score was therefore used to quantify the level of artifacts in the signals during the running and immobility phases (see Supplementary Table 1).
Statistics
The number of animals was estimated by a Monte Carlo simulation using R software (R Core Team 2017). According to the test, for a five-speed exercise protocol, a group of eight animals offered a power of 99% and 97% to detect an increase in HR and MV, respectively, for each speed level (analysis of variance (ANOVA) test). Given the animals likely to refuse to run on the treadmill 24 and other potential protocol issues, two animals were added to this estimate to ensure sufficient statistical power, resulting in a total of 10 animals.
Statistical analyses were performed with Python scripts using the Pinguin statistical package (v0.5.1). 25 The normality of the data distribution for each speed level was verified using a Kolmogorov–Smirnov test. A one-way repeated measures ANOVA was calculated to follow the evolution of HR, MV, and RespR during exercise and recovery situations. Differences in HR, MV, and RespR from the resting reference condition were compared using the Student’s paired t-test. The limit of statistical significance was p < 0.05.
Results
Equipment with the jacket and running on the treadmill
All animals accepted the jacket after completion of the habituation phase. Placing them back in their cage after the equipment for the pre-exercise stabilization was associated with an initial sustained AL of 152 ± 17 mgn. AL significantly decreased to 8 ± 1 mgn during the 1 h-stabilization (p < 0.001). This was associated with a significant reduction in HR (from 488 ± 9 to 345 ± 5 bpm, p < 0.001), MV (from 1618 ± 191 to 751 ± 70 ml min−1 kg−1, p < 0.01) and RespR (from 261 ± 7 to 115 ± 6 cpm, p < 0.001).
The jacket did not impede movement during the treadmill exercise (Figure 1). Out of 10 animals, 9 completed the speed levels and durations (speed level maximum 40 cm s−1 and running duration 25 min) expected in the exercise protocol without showing any abnormal locomotion patterns due to the device. Although all 10 animals reached maximum speed during the habituation phase, one did not complete the last level of the protocol (40 cm s−1) and was excluded from the analysis of this level (n = 9 for 40 cm s−1).

Wistar rat equipped with the monitoring jacket during exercise on the single-track treadmill (e.g., Rat # 4, treadmill speed level = 20 cm s−1).
Running induced a significant increase (p < 0.05) in AL (300 mgn; +227%) between the 5 and 40 cm s−1 levels. AL was positively correlated to the speed level with a Pearson correlation coefficient of 0.973. The average locomotor AL measured during break was 52 ± 9 mgn, consistent with animals moving moderately and intermittently.
Physiological monitoring and detection scores
HR, MV, and AL increased progressively with treadmill speed levels, whether they were measured during the running phases or during breaks (Figures 2 and 3). Each break (B) induced an immediate decrease in ventilatory variables when the treadmill stopped (Figure 2).

Screenshot of the software (Rat #2). Time course of ventilatory variables during exercise presented as means ± SEM every 5 s over the whole exercise protocol. Heart rate (HR in bpm), minute volume (MV in ml min−1) and locomotor activity level (AL in mgn). Speed level markers (dark arrows) were positioned at the beginning of each corresponding speed level. B markers (B = break) were positioned at the end of each speed level to identify the beginning of the 30-s immobility phase.

Evolution of (a, b) cardiac detection score (solid red line) and respiratory detection score (blue dotted line) (in %); (c, d) heart rate (HR in bpm); (e, f) minute volume (MV in ml min−1 kg−1); and (g, h) respiratory rate (RespR in cpm), recorded by the jacket during treadmill exercise. Cardiorespiratory variables were measured at break (a, c, e, g) and run (b, d, f, h) conditions at the end of the speed level. Values are reported as mean ± SEM for each speed from baseline (BL) to 40 cm s−1 speed with n = 10 animals (except for 40 cm s−1 where n = 9). *p < 0.05, **p < 0.01, and ***p < 0.001 for post hoc paired Student’s t-test with ANOVA p < 0.005 for HR and p < 0.001 for MV.
The detection score of ECG at BL was 96% on average and remained stable throughout the protocol, in both stationary (Figure 3a) and running (Figure 3b) conditions as speed increased, indicating that running did not affect cardiac cycle detection. The ventilatory cycle detection score diminished from 97% at rest (BL) to 62% after starting the treadmill and gradually decreased with the speed before stabilizing at around 40% for speeds of 20 cm s−1 or higher. Despite the impact of exercise movements on the detection of respiratory cycles, the evolution profile of MV and RespR were equivalent during running and stationary (break) periods.
Cardiorespiratory response to exercise
Exercise on the treadmill induced an increase in HR (Figure 3c and 3d), MV (Figure 3e and 3f), and RespR (Figure 3g and 3h). The increase was statistically significant as confirmed by the one-way ANOVA and subsequent post hoc paired t-tests under both run and break conditions. HR at exercise (Figure 3a) significantly increased compared to BL (+99 bpm, at exercise, p < 0.001), followed by a more discrete increase for subsequent levels. The overall increase in HR at the last speed level compared with BL was +178 bpm. Ventilatory adaptation presented an initial MV increase of 1271 ml min−1 kg−1 at exercise (p < 0.01). The overall increase at the last speed level compared with BL was +2234 ml min−1 kg−1. RespR followed a similar trend with an overall increase of 164 cpm.
Figure 4 compares the evolution of HR, MV, and RespR after the termination of the exercise with the initial BL condition. The ANOVA showed a significant recovery over time for HR (p < 0.001), MV (p < 0.001), and RespR (p < 0.001). The recovery dynamics of HR, MV, and RespR were rapid in the first 40 min after exercise and then became non-statistically significant over 10-min intervals thereafter.

Post-exercise recovery profile monitored by the jacket. Time course of (a) heart rate (HR in bpm), (b) minute volume (MV in ml min−1 kg−1), and (c) respiratory rate (RespR in cpm) recorded during 60 min post-treadmill exercise and compared to pre-exercise baseline (BL). Values are presented as means ± SEM calculated over 30-s recording segments for each time point with n = 10 animals. *p < 0.05, **p < 0.01, and ***p < 0.001, for paired Student’s t-test comparing BL with each time point.
Discussion
This study demonstrated that laboratory rats can perform a standardized running exercise without being hindered by the DECRO jacket and associated on-board equipment. In addition, the performance of the device was suitable for monitoring relevant cardiorespiratory features during exercise and subsequent recovery, despite the motion artifacts induced by running on the treadmill.
In the present work, fitting the animals with DECRO was followed by a 60 min habituation and stabilization period, as initially described by Fares et al.20,21 HR measured after the stabilization period prior to exercise fell within the range reported for rats monitored under similar conditions.26 –28 Likewise, BL MV values were comparable to data published in the literature for resting rats.27,29
In the context of an incremental speed exercise protocol, it was necessary to evaluate whether the animals were disturbed by the jacket or the weight of the electronic device located on their back. From an observational point of view, our results showed that all animals could run when equipped with the device and were able to reach the 40 cm s−1 speed level. In addition, the evolution of HR during the exercise protocol in our study was perfectly comparable to that reported by Kuckmarski et al. for the same exercise levels (resting HR = 350 bpm; HR at 41.6 cm s−1 = 550 bpm). 7 This observation further confirmed the absence of influence of our device on the physiological parameters measured during exercise.
Several previous studies have already demonstrated the effectiveness of DECRO for the cardiovascular monitoring of unrestrained laboratory rats during their daily phases of rest and activity.20,21 Nonetheless, its ability to accurately detect signals during high-intensity movements remained unexplored. In this study, motion artifacts were indirectly quantified using the detection score provided by the software. Each excluded ventilatory or cardiac cycle reduces the detection score, so the lower the score, the greater the impact of the movements. As evidenced by the ECG detection score, HR calculation was proved to be highly insensitive to motion artifacts and remained close to 100% throughout the protocol. Respiratory variables were found to be more sensitive to running-induced motion artifacts with a decrease in detection score from 97% (at BL) to 39% (at a speed of 30 cm s−1). Nevertheless, since exercise is associated with an increase in RespR, the number of analyzable cycles per 30-s analysis window remained almost constant regardless of the level of exercise (55 cycles on a total of 56 at rest and 50 on a total of 119 at a speed of 20 cm s−1). In addition, comparison of the data acquired with and without motion artifacts was essential to better assess the extent to which they may influence physiological outcomes. To enable this comparison we adapted the design of the exercise protocol by introducing short periods of immobility (breaks). Given their very short duration, such breaks were not expected to allow a significant cardiorespiratory recovery. 30 The values measured during these breaks can therefore be considered representative of the value at the end of the previous exercise level. Since the profiles of HR, RespR, and MV were equivalent during exercise and breaks, the ability of the device to acquire cardiorespiratory data under both conditions was confirmed.
Adaptation of cardiac function to exercise was assessed using HR changes as a marker. The animals showed a response consistent with that reported in previous studies involving animals of similar age and weight monitored using implantable ECG recording devices.28,31,32 This response was characterized by a sharp increase in HR at the start of exercise, followed by a slower, gradual increase as a function of running speed, tending toward a plateau corresponding to maximum HR. In addition, the evolution of HR during post-exercise recovery followed a typical pattern including a rapid initial decrease followed by a slower phase. 33 The acquisition of MV by the device enabled direct monitoring of ventilatory adaptation to exercise, which is very difficult to implement in rodents using conventional equipment. 2 In our experiments, the first exercise step initially induced a sharp increase in MV followed by a more gradual increase for subsequent stages. From a speed of 30 cm s−1, MV reached a plateau, resulting in an overall increase of 297% compared with BL values. To date, no direct MV measurement during exercise has been described in rats and therefore there are no published data that would allow us to verify the validity of our MV results on rats subjected to exercise conditions. Nonetheless, it is accepted in the literature that VO2 (i.e., oxygen consumption) can be considered a variable that, in humans, follows the evolution of MV. 31 A previous study on rats undergoing a treadmill exercise protocol similar to ours measured the VO2 of the animals at different exercise levels. 32 The VO2 evolution profile reported in this study is perfectly superimposable on that we obtained with MV, thereby providing an indirect validation argument for our ability to monitor this variable.
Limitations of the study
Our study showed that the DECRO system makes it possible to collect scientifically relevant cardiorespiratory data from laboratory rats in a controlled exercise situation, using a totally non-invasive approach. However, although non-invasive, the device is not unobtrusive and the fact of wearing the equipment, even if it leaves the rat free to move, constitutes an unnatural situation that is likely to have an impact on its behavior. Further studies are needed to assess this impact, particularly before considering the use of DECRO in behavioral studies. Another major issue in rodent treadmill exercise protocols is the choice of the negative stimulation system. For laboratory rodents, treadmill running is always a stressful exercise requiring negative reinforcement, at least during the habituation phase. The two methods most widely used are light electric shocks and short air puffs delivered by a compressor. To enable an ethically justified choice, an earlier study compared the level of stress induced by these two methods by assessing different biomarkers such as corticosterone and adrenocorticotropic hormone, and no difference was observed, suggesting that the two techniques are equivalent in terms of the harm imposed on the animals. 34 In the present work, we chose the electric shock technique, as recommended by Adamovich et al., 35 because our treadmill is equipped with an electric stimulation grid, like the majority of commercially available treadmills. In addition, the mild electric shock technique is highly effective and does not interfere with the running exercise, unlike other methods involving mechanical stimulation of the animal. 35 Even if there is no a priori reason for the DECRO system to be incompatible with any of the negative reinforcement techniques available on the market, it will be necessary for each method to check that it does not cause the animal any particular discomfort due to the presence of the equipment, and that it does not interfere with the measurements.
Finally, as one of our concerns was to carry out our study using as few animals as possible, we only used naive adult male rats. Of course, the validity of the system in other potentially relevant preclinical use cases, for example in female rats or in pathological, obese, or senescent animals, will also have to be established beforehand.
Conclusion
Although various technologies have been proposed for the non-invasive acquisition of cardiac and respiratory11 –14 activity data in laboratory rodents, no device currently available on the market, either under development or at the proof-of-concept stage, provides simultaneous access to these two types of data, particularly under conditions of movement. A previous study 21 has demonstrated the absence of effect of the DECRO device on animals at rest, after 3 days of habituation, on a battery of 58 observations covering awareness, motor activity, and behavioral changes over six consecutive hours. In addition to this, our study further demonstrated the ability of jacketed telemetry to provide non-invasive monitoring of cardiac and respiratory functions in laboratory rats during treadmill running exercise. It should be noted that the electrocardiographic acquisition system of the DECRO device, which was only used in this proof-of-concept study for the detection of R-peaks, could also be a considerable asset for the study of cardiac rhythm disorders in certain preclinical situations. Finally, the device made it possible to monitor functional variables that were previously inaccessible under such conditions. In conclusion, the DECRO device enabled robust assessment of cardiorespiratory adaptation to exercise in laboratory rats. The preclinical applications of this device are numerous, notably for the study of metabolic, cardiac, and respiratory diseases, or for assessing the efficacy of pharmacological compounds in healthy and pathophysiological models.
Supplemental Material
sj-pdf-1-lan-10.1177_00236772241259857 - Supplemental material for Jacketed telemetry in rats: a novel non-invasive method for cardiorespiratory phenotyping during treadmill exercise
Supplemental material, sj-pdf-1-lan-10.1177_00236772241259857 for Jacketed telemetry in rats: a novel non-invasive method for cardiorespiratory phenotyping during treadmill exercise by Stéphane Tanguy, Agathe Cambier, Leandro Fontana-Pires, Timothé Flenet, Charles Eynard, Julie Fontecave-Jalon, Pierre-Yves Gumery and François Boucher in Laboratory Animals
Footnotes
Data availability statement
Detailed data are available as supplemental data, simultaneously submitted with the paper.
Parties interested in accessing the data can contact the authors at one of the two following email addresses:
Declaration of conflicting interests
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: ETISENSE supported this study by lending the research team the DECRO system and providing the connected jackets.
Supplementary material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
