ABSTRACT
Captive housed non-human primates, specifically great apes such as chimpanzees (Pan troglodytes) are frequently reported to have died from or are diagnosed with potentially fatal heart conditions that require the monitoring of physiological signals such as electrocardiogram (ECG) or respiratory rate. ECG screening must be conducted after applying full anaesthesia, causing potential physical and emotional stress as well as risk for the animal. Here, we present an electronic system that simultaneously measures the ECG and the electrical bioimpedance for the early detection of abnormal cardiovascular activity. Modified gloves whose fingers are equipped with electrodes enable the caregiver to obtain three cardiovascular signals (ECG, pulse rate and respiratory rate) by placing the fingertips on specific parts of the non-human primate without needing any prior physical preparations. Validation (ECG and bioimpedance) was performed both on humans and on captive housed chimpanzees, where all the signals of interest were correctly acquired.
INTRODUCTION
Heart failure leading to a sudden cardiac death have been reported as the most common cause of death in captive housed chimpanzees (Baldessari et al., 2013; Drane et al., 2019, 2020; Lammey et al., 2008a,b; Varki et al., 2009), and in other great apes and monkeys as well (Murphy et al., 2018). The most recent study reviewing spontaneous causes of mortality in captive chimpanzees, by Laurence et al. (2017), reported that lesions involving the cardiovascular system (51.1%) caused more than half of all mortalities. These heart-related mortalities could be subdivided into lesions that in 90% of cases affected the heart itself and in 10% affected the vasculature. More specifically, 87.3% of all heart-related mortalities were diagnosed as chronic cardiomyopathy and this was reported to occur four times more often in males than in females (Laurence et al., 2017). Some heart diseases may produce specific cardiac remodelling, which may give rise to an increased risk of cardiac events leading to sudden death (Payne et al., 2015; Pfeffer and Pfeffer, 1987), often associated with myocardial fibrosis (Magden et al., 2016; Strong et al., 2020), which increases the risk of cardiac arrhythmias (Magden et al., 2016). Nevertheless, by characterizing the cardiac function in aging chimpanzees – detecting heart diseases at early stages (Baldessari et al., 2013) – some cardiac deficiencies can be reversed or at least ameliorated (van Zijll Langhout et al., 2017) by adapting care strategies and using pharmacological agents (Fagard et al., 2009; Pfeffer et al., 1985; Sleeper et al., 2005).
However, the early detection of cardiovascular deficiencies in great apes is challenging, as it typically requires the animal to undergo frequent health checkups under full anaesthesia, which are costly and risky events, and place a physiological and emotional burden on the animals (invasive procedures). Consequently, diseases related to the heart, such as abnormal heart rhythms (Lowenstine et al., 2016), are often detected once the disease has already progressed and behavioural indicators or secondary effects become apparent.
Echocardiography and magnetic resonance imaging are the most efficient methods to detect abnormalities in the cardiovascular system. Yet, this is not always possible because of the lack of resources and, at times, the remote location of institutions taking care of chimpanzees in captivity (Drane et al., 2020). Besides the captive population at zoos, there are several primate rescue centres and sanctuaries, such as the Tchimpounga Chimpanzee Rehabilitation Centre (Tchimpounga, Republic of Congo) or the MONA Sanctuary (Girona, Spain). There, primates are cared for after being rescued from illegal and harmful conditions, typically having experienced traumatic events and adverse living conditions (Crailsheim et al., 2020; Feliu et al., 2022; Hevesi, 2023; Kalcher et al., 2008). Most of these rescued primates arrive with psychological, emotional and often serious physical deficiencies including respiratory and/or cardiac conditions that require professional treatment and constant monitoring.
In humans and in several other species of non-human animals, diagnostic criteria serving as potential indicators of adverse cardiac remodelling have been successfully developed by studying the relationship between electrocardiogram (ECG) and measures of cardiac structures (Aro and Chugh, 2016; Hancock et al., 2009; Moise et al., 1986). More recently, in an effort by Drane et al. (2020), a large heterogeneous sample of chimpanzees (zoo-housed and African-sanctuary chimpanzees) was examined in order to gain insights regarding results for ECG variables and for echocardiographic imaging, which would, in continuation, allow the development of ECG prediction equations that potentially provide veterinarians with an effective tool to assess cardiac structures and detect heart abnormalities in chimpanzees.
Beyond the ECG, in veterinary practice or in animal testing in biomedical research (Detweiler, 2010; Erickson and Olsen, 1985), it is common to monitor the respiratory rate (RR) or the blood pressure (BP), which are used to derive the heart rate (fH) or the pulse rate (PR). Besides healthcare, such signals are important to study the physiology of animals (Ibbini et al., 2022). For example, fH has been proposed as an indirect predictor of the metabolic rate of some animal species under specific conditions (McPhee et al., 2003; Portugal et al., 2009).
Various invasive and non-invasive techniques exist (do Nascimento et al., 2021; Ho et al., 2011; Konopelski and Ufnal, 2016) that yield electrocardiographic signals with different levels of stress on the test subjects. The Holter monitor is commonly used to track the ECG in animals (Lammey et al., 2011), yet chimpanzees do not cooperate throughout such measurement procedures and typically must be anesthetized (Atencia et al., 2015; Botelho et al., 2019; McPhee et al., 2003). More invasive – and common – approaches include the implantation of ECG recording electrodes (Portugal et al., 2009; Sriram et al., 2021) and intravenous cannulation (Lin et al., 2019). The latter is frequently used in smaller animals, such as rodents. BP usually needs an inflatable cuff and is usually measured on the tails of monkeys and rodents (Chester et al., 1992; Reddy et al., 2003), causing different levels of movement restriction and alterations in their behaviour. This is not possible for monkeys with vestigial tails, hence they are anesthetized and BP is measured on the arms with a sphygmomanometer or a cuff oscillometer (Brownlee et al., 2020). Implantable loop recorders have been successfully used in chimpanzees diagnosed with ventricular premature complexes to monitor cardiac arrhythmia (Lammey et al., 2011; Magden et al., 2016). fH can also be computed from the ballistocardiogram, a non-invasive technique that measures the force exerted by the pumping of arterial blood (Xie et al., 2019), and that has been successfully used to detect fH in whales (Czapanskiy et al., 2022). In other large animals, straps or garments embedded with the required sensors are frequently used (Antink et al., 2019; Brugarolas et al., 2014).
Electrical bioimpedance (BI) is a measurement technique enabling us to determine the body composition of an individual, e.g. assessing the amount of water of a living being (Matthie et al., 1998; Smith et al., 2009; Ward et al., 2016). BI also permits the tracking of physiological changes such as the respiratory and cardiac (heartbeat) activity (Buchan and Tublitz, 1988; Lee and Cho, 2015). This is known as impedance plethysmography (IPG). Despite its potential to provide information regarding relevant cardiorespiratory parameters, it is seldom used in animal studies because of the problem of electrode attachment (Anand et al., 2021). Moreover, there is more cardiovascular information that can be obtained by measuring the pulse arrival time (PAT) or the pulse transit time (PTT), defined as the time delay between the ECG and the IPG at proximal or distal sites, respectively (Gomez-Clapers et al., 2015). Thus, acquiring both signals simultaneously broadens the available tools to study the physiology and the health status of animals. Either in ECG or in IPG studies, keeping a small electrode-contact impedance (Ze) without electrode implantation is a major challenge. In a study by Yang et al. (2022), Ze in rat skin – with hair shaved down to 0.8 mm – was at least an order in magnitude larger than in human skin in the ECG bandwidth (a few hundreds of hertz). BI works with kilohertz-range frequencies, where Ze is much lower, and hence is less of an inconvenience. Nevertheless, the mechanical attachment remains a problem.
Here, we present a simultaneous ECG and IPG measurement system that enables us to measure the cardiovascular condition of large captive non-human primates, using concealed electrodes. Specifically, we have modified a pair of protection gloves by covering their fingers with conductive material, converting them to effective electrodes that will go unnoticed by the animal. Animal care staff who routinely establish physical contact with animals, i.e. caregivers and/or veterinarians, can wear these gloves to obtain measurements by placing the electrodes via the fingertips of the glove on the animal's body. Therefore, ECG and IPG measurements can be obtained frequently and fast, without the need of prior physical manipulation, excessive movement restriction or anaesthesia. Furthermore, some pharmaceutical agents used for anaesthesia were reported to alter values of ECG readings (Magden et al., 2016). This novel measurement system has the potential to serve as an early detection system of abnormal health conditions in great apes, or for tracking the progression of cardiac and/or respiratory illnesses and treatments.
MATERIALS AND METHODS
Glove electrodes
Conductive fabric caps were sewn on to cover the fingers of a glove (Fig. 1A), converting each fingertip into a dry electrode capable of measuring ECG and IPG signals. The principal objective of this design was to hide and protect sensing cables as well as to ensure that the animals would not be alarmed or become wary of excessive equipment, ideally not being aware of the measurement process at all. Hence, these modified gloves become the measurement interface between the animal and the electronic instrument. The gloves were equipped with an internal double electrical insulation to ensure that the user, i.e. caregiver or veterinarian, does not contribute to the measured signals. All five fingers on both gloves were modified to act as active elements of the measurement (10 electrodes in total), making them easily reconfigurable, yet only four are required to be active in order to achieve the desired measurements (two electrodes on each hand). Fig. 1B shows the electrical connections of the electrodes to the electronic instrument. For the wiring connecting the electrodes of the gloves to the electronic instrument, we used thin cables that ran along the user's arms to ensure that animals could neither see nor get hold of any cables.
ECG and IPG instrumentation
In Fig. 1C, we graphically depict how electrodes A and B inject the low-level current necessary to perform the IPG measurements, while electrodes C and D pick up the resulting IPG signal response (four-electrode technique) and the ECG signal. Because of the high Ze expected in animals, electrodes C and D are connected to operational amplifier (op-amp) voltage buffers at the gloves, becoming active electrodes that avoid voltage loading but also the errors associated with the long (1 m) connection cables necessary to hide the electronic instrument from the animals. The aim was to obtain both the ECG and the IPG signals with a high enough signal-to-noise ratio (SNR) as to allow for posterior simple signal processing methods to calculate RR, fH and PAT. Furthermore, Fig. 1C shows the block diagram of the electronic system implemented to acquire the two signals simultaneously. The upper branch of Fig. 1C shows the ECG signal-conditioning channel, and the lower branch, the IPG signal-conditioning channel. All of the subsystems were supplied at ±5 V.
Current source
The circuit generates a 20 kHz, 0.20 mA (root mean square, rms) current by means of a Howland current pump and a Wien oscillator. This current is injected between electrodes A and B (Fig. 1C). Electrode B is connected to ground. Previous measurements in humans showed that the pulsatile impedance variations when measuring between both hands were about 400 mΩ, equivalent to a voltage response of the order of 0.1 mV. Therefore, a gain of approximately 10,000 was set to amplify it up to a 1 V amplitude.
IPG signal conditioning
In the IPG channel, a band-pass filter from 1 kHz to 100 kHz eliminates the ECG signal and limits the input noise. The IPG signal is amplified (G2≈5) and then demodulated with a +1/−1 switched-gain amplifier circuit (Gomez-Clapers et al., 2015). The resulting voltage is proportional to the bulk tissue impedance, but also contains changes due to respiration and heartbeats – three orders of magnitude lower. Hence, the basal impedance has been filtered out with a 0.5 Hz high-pass filter and the low-level physiological variations are largely amplified (G3≈2000) and limited to a bandwidth of 20 Hz by a low-pass filter stage.
ECG signal conditioning
Besides electrodes C and D, electrode B was used as the ground electrode. The IPG signal was filtered out by a passive first-order fully-differential band-pass filter with cut-off frequencies of 0.5 Hz and 100 Hz, recovering only the ECG signal. This signal was amplified with a 1000-gain instrumentation amplifier, and the output noise bandwidth was limited to 40 Hz by the ensuing passive first-order low-pass filter.
Digital signal acquisition
A USB-231-OEM DAQ (Measurement Computing), which has 16-bit resolution, digitized both the IPG and the ECG signals with a 1 kHz sampling frequency. This data acquisition (DAQ) system was connected to a laptop PC where a LabVIEW (National Instruments, Inc.) application received and stored the digitized data. Afterwards, the data were processed with Matlab (MathWorks Inc.).
System validation
Preliminary ECG and IPG measurements were performed on seven human volunteers, six men and one woman, aged 41.3±15.1 years. These signals were obtained simultaneously with both the developed system and a commercial MP36 system (Biopac Systems, Ltd). For the IPG measurements, the MP36 was configured for impedance cardiography (ICG) measurements as it was the closest available in that kind of system. The volunteers breathed normally for a few seconds, while in a sitting position during the measurements. The data collection in this group of volunteers adhered to the ethical considerations of the World Medical Association Declaration of Helsinki, and adhered to the legal requirements of Spain. The volunteers gave their written informed consent.
Measurements in chimpanzees
A set of measurements were performed on four male chimpanzees [two Pan troglodytes (Blumenbach 1775) aged 34 and 36 years, and two hybrids from Pan troglodytes and Pan troglodytes verus aged 20 and 23 years) at the MONA Sanctuary. As shown in Fig. 1D–F, the caregiver wore the designed gloves to obtain the measurements. The output of the proposed measurement system was connected to the data acquisition laptop via a 10 m long active USB cable. The measurement was carried out in the MONA sanctuary's socialization cages, which include an extra attached cage that allows for close physical examination (veterinary purposes), and is thus the optimal location to perform these measurements (Fig. 1F). The caregiver held the hands of the chimpanzees at rest and, based on his experience, he reassured the animals in order to obtain measurement records of at least 30 s duration, the minimum time required for the acquisition of the ECG and the IPG signals. Such ultra-short-time (UST) acquisition methods produce results with a high correlation with the gold-standard 5 min measurements (Schäfer and Vagedes, 2013). The caregiver also tested other body parts, where he could safely touch the chimpanzee, yet the best signal quality was obtained when touching the hands. Because this test was performed on non-anaesthetized subjects, it was not possible to use a standard electrocardiograph simultaneously to the gloves; hence, validation of the acquired data was done by contrasting against the ECG reference intervals reported in Atencia et al. (2015).
No animals were forced to participate in the data collection of this study. All the research strictly adhered to the legal requirements of Spain, and followed the institutional guidelines for the care and management of primates as established by the MONA Sanctuary.
Data processing
The Pan–Tompkins algorithm was applied to the chimpanzees’ ECGs to detect the R wave, allowing calculation of the tachogram, used in fH variability studies. Erroneous detection of R waves from heartbeats contaminated by movement artifacts was avoided by adding a temporal threshold of maximum R–R (time interval between consecutive R waves) time variation of 40% between one beat and the next to the detection algorithm. To calculate the breathing rate, the IPG signals were digitally band-pass filtered from 0.06 Hz to 0.5 Hz by a 3rd order Butterworth high-pass infinite impulse response (IIR) filter and a 3rd order Butterworth low-pass IIR filter. Filtering was done using the forward–backward technique to avoid any signal distortion that would have otherwise been caused by the non-linear phase response of the IIR filter. As in humans, the time interval between the ECG and the IPG allows pulse wave velocity (PWV) and the PAT to be obtained. Following the methods in Gomez-Clapers et al. (2015), the arrival of the blood pressure pulse was determined from the timing of the foot of the systolic wave identified by the intersecting tangents method. Therefore, the PAT was calculated as the time interval between the detected R wave and the IPG systolic foot.
RESULTS AND DISCUSSION
Validation in humans
The raw ECG signals obtained using our electronic instrument and the MP36 did not differ, and the IPG and ICG signals showed the same behaviour (see Fig. 2). When post-processed, a correlation higher than 99% was obtained between the ECG and the IPG signals.
Chimpanzees’ ECG and IPG measurements
Fig. 3A shows the ECG measurement of lead I (Fig. 1E) obtained on a chimpanzee during approximately 50 s. Although there was some electromyogram (EMG) contamination and small movement artifacts, it was possible to detect the animal's fH without difficulty. After processing (Fig. 3B), the R waves were detected by the Pan–Thompkins algorithm. Fig. 3D shows the resulting fH tachogram for this chimpanzee. In this recorded signal, only one erroneous beat was detected due to a motion artifact that was effectively eliminated by the R–R temporal threshold.
The UST tachogram enabled fH variability analysis, showing a mean fH of around 65 beats min-1. The QRS interval could be identified, showing a value around 80 ms. The P and T waves were visible but not well defined, which is usual in lead I ECG recordings. This is in agreement with the reference values reported in Atencia et al. (2015). Although the chimpanzee was not agitated and did not show any indication of being nervous/stressed, a relatively large variability was observed. This might be an effect of the caregiver touching the subject, as discussed later.
Focusing on a short time period in Fig. 3C, it is possible to see the ECG (blue) and the IPG (red) details. The IPG shows a large influence of respiration. This allowed computing of a RR of 14.6 breaths min−1. Because IPG is directly related to the BP wave, the PR might also be computed from it and could be considered a fH substitute in future studies. Finally, a PAT of 80 ms was calculated (Fig. 3E). This is much shorter than reported PAT values in humans (around 200 ms), but no reference data exist in chimpanzees to contrast this finding. All of these parameters were measured at distal sites, which avoids the need for proximal – chest – sites, which are usually problematic (high electrode impedance) because of fur and the physiology of the animal. Nonetheless, the buffered electrodes should withstand these issues. Therefore, the system could also be used at proximal sites in trained animals that are not disturbed by the presence and the touch of their caregivers.
Benefits and limitations
These results show the feasibility of using the gloves to obtain cardiovascular measurements in great ape species. The designed system is an improvement over current methods as it allows the most important cardiovascular parameters of a chimpanzee to be obtained at high quality and with minimal processing, without the need to physically prepare the animal or use invasive procedures such as anaesthesia. The low cost of this instrumental solution and its relatively easy application enables primate housing institutions (irrespective of financial background) and their trained care staff to effortlessly incorporate ECG and IPG measurements into the animal care management on a regular basis. If, as suggested, this system is incorporated and frequently used, especially in male chimpanzees older than 20 years, it has the potential to serve as an early detection system of abnormal cardiac health conditions without the need for risky, costly, time-consuming and potentially stressful invasive events. Furthermore, veterinarians have another efficient tool to effectively monitor the progression of cardiac and/or respiratory illnesses as well as confirm the effectiveness of treatments in animals that, because of cardiac or respiratory impairments, are at risk when put under anaesthesia.
There are, nevertheless, some limitations to the applicability of these gloves. The gloves go unnoticed by the animal, but the presence of the caregiver in itself might induce fH variability. This is not to say that his or her presence is perceived as negative or stressful, but it is indeed a potentially influencing factor, as per Grandi and Ishida (2015). In that study, two captive male rhesus monkeys (Macaca mulatta) were groomed by a familiar caregiver/experimenter – not visible to the subjects – while their ECGs were acquired, and this action notably decreased fH and, at the same time, increased fH variability. In the present study, the caregiver touched the animal and kept eye contact, which might explain the chimpanzee's large fH variability (Fig. 3C). However, this influencing impact might be greatly reduced by a repeated and uniform application of the measuring system, converting this animal–caregiver interaction into a routine activity, i.e. reducing the time needed to achieve the measurements and reducing the potential arousal effect (either negative or positive).
In contrast to a full electrocardiographic diagnosis system, our system is intended to be used as a fast screening and/or health-monitoring tool, hence a single lead is deemed sufficient. A follow-up study will provide more data on the usefulness of such a single ECG-lead system.
Conclusions
Here, we present a new system for measuring cardiovascular parameters in captive chimpanzees. The system was validated in humans and tested in chimpanzees, yielding the expected physiological information. Care staff can incorporate the use of the modified gloves into their daily routine within the basic animal management procedures, enabling fast and periodical measurements without the need for anaesthesia or complex preparations. The simultaneous acquisition of the ECG and IPG signals enables us to deepen our currently incomplete understanding regarding the cardiovascular physiology of chimpanzees, as currently done in state-of-the-art human applications. Because of the good signal quality acquired for periods of more than 30 s, in addition to the waveforms themselves, it is possible to obtain information on RR, fH and its variability, as well as PAT and PWV values. This simple system can also be useful in other captive-housed wild animals, where cardiac screening might be necessary.
Acknowledgements
The authors thank the staff at the MONA Sanctuary for their assistance in the performance of the measurements.
Footnotes
Author contributions
Conceptualization: E.S.-F., G.H., F.L., D.C., O.F., O.C.; Software: O.C.; Validation: G.H., S.M., O.C.; Formal analysis: E.S.-F., G.H., O.C.; Investigation: E.S.-F., G.H., D.C., O.C.; Resources: O.F., O.C.; Writing - original draft: E.S.-F., S.M., O.C.; Writing - review & editing: G.H., D.C., O.F.; Supervision: O.F., O.C.
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Data availability
All relevant data can be found within the article.
References
Competing interests
The authors declare no competing or financial interests.