If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Safety evaluation of smart scales, smart watches, and smart rings with bioimpedance technology shows evidence of potential interference in cardiac implantable electronic devices
Department of Cardiology, Heart Institute, Hospital Universitari Germans Trias i Pujol, Badalona, SpainCentro de Investigación Biomédica en Red Enfermedades Cardiovasculares (CIBERCV), Madrid, SpainAutonomous University of Barcelona, Barcelona, Spain
Address reprint requests and correspondence: Dr Benjamin Sanchez, Sorenson Molecular Biotechnology Building, 36 S Wasatch Drive, Office 3729, Salt Lake City, UT 84112.
Smart scales, smart watches, and smart rings with bioimpedance technology may create interference in patients with cardiac implantable electronic devices (CIEDs).
Objectives
The purpose of this study was to determine interference at CIEDs with simulations and benchtop testing, and to compare the results with maximum values defined in the ISO 14117 electromagnetic interference standard for these devices.
Methods
The interference at pacing electrodes was determined by simulations on a male and a female computable model. A benchtop evaluation of representative CIEDs from 3 different manufacturers as specified in the ISO 14117 standard also was performed.
Results
Simulations showed evidence of interference with voltage values exceeding threshold values defined in the ISO 14117 standard. The level of interference varied with the frequency and amplitude of the bioimpedance signal, and between male and female models. The level of interference generated with smart scale and smart rings simulations was lower than with smart watches. Across device manufacturers, generators demonstrated susceptibility to oversensing and pacing inhibition at different signal amplitudes and frequencies.
Conclusions
This study evaluated the safety of smart scales, smart watches, and smart rings with bioimpedance technology via simulation and testing. Our results indicate that these consumer electronic devices could interfere in patients with CIEDs. The present findings do not recommend the use of these devices in this population due to potential interference.
Graphical abstract
Graphical AbstractSmart scales, smart watches, and smart rings with bioimpedance sensing technology have the potential to interfere with the correct functioning of cardiac implantable electronic devices.
Medical bioimpedance technology, until recently only available in hospitals, is now quickly spreading into consumer and wearables devices as companies increasingly incorporate bioimpedance sensing capabilities into their new electronic products. Driven by the increasing demand of consumers to monitor their own health and the benefits of doing so continuously outside of the hospital setting, bioimpedance technology is becoming more ubiquitous. Examples of mainstream consumer electronic products commercially available at home include body composition smart scales (eg, Fitbit Aria 2, San Francisco, CA and Withings Body Scan, Issy-les-Moulineaux, France), smart watches (eg, Samsung Galaxy Watch 4, Suwon-si, South Korea and Empatica E4, Boston, MA), and smart rings (eg, Moodmetric, Kouvola, Finland).
The underlying principle of bioimpedance consists of applying via 2 electrodes an alternating, low-amplitude, painless electrical current and then measuring the resulting voltage generated by the body using a different pair of electrodes. Changes that occur as a result of disease will alter the ionic and cellular integrity of tissues and fluids, thus affecting their ability to conduct alternating electrical current, which ultimately will impact the characteristics of the voltage signal recorded. Sweeping through a frequency range of interest, the voltage-current waveforms are used to calculate the electrical impedance of the body (ie, bioimpedance) using Ohm’s law. Simply expressed, impedance equals voltage divided by current,
Single-frequency BIA devices to estimate body composition parameters such as total body water typically measure at 50 kHz, whereas multifrequency devices measure a larger number of frequencies in the kilohertz to megahertz range. Noninvasive hemodynamic ICG monitoring is based on the measurement of thoracic bioimpedance between 50 to 100 kHz. Beyond BIA and ICG, measuring bioimpedance has shown clinical value in patients with neuromuscular disorders
Analysis of different device-based intrathoracic impedance vectors for detection of heart failure events (from the Detect Fluid Early from Intrathoracic Impedance Monitoring Study).
Despite the clinical value, no consumer bioimpedance device has been cleared by the US Food and Drug Administration (FDA) for subjects with cardiac implantable electronic devices (CIEDs) due to possible electrical interference. For example, manufacturers of the Samsung Galaxy Watch 4 and both the Fitbit Aria 2 and the Withings Body Cardio smart scales have disclaimers on their Web sites preventing the use of their devices by subjects with implanted electronic medical devices.
Over the last few years, few studies on electrical interference of bioimpedance technology on CIEDs have been reported.
To date there is a lack of publicly available benchmark data to close the gap in knowledge from these observational studies using medical instrumentation about the extent to which CIEDs may be influenced by bioimpedance technology available in consumer electronic devices.
In this study, we evaluated the electrical safety of measuring bioimpedance using technical specifications from the FDA-recognized ISO 14117.
ANSI/AAMI/ISO 14117, Active implantable medical devices—Electromagnetic compatibility—EMC test protocols for implantable cardiac pacemakers, implantable cardioverter defibrillators and cardiac resynchronization devices, 2019. https://www.iso.org/standard/73915.html. Accessed December 21, 2022.
This standard defines reproducible benchtop test methodologies and voltage threshold values that manufacturers of CIEDs must use to demonstrate that their devices achieve the appropriate level of electromagnetic compatibility in uncontrolled electromagnetic environments that patients with these devices might encounter. The purpose of this study was to determine the level of interference in CIEDs during bioimpedance (1) simulations and (2) benchtop testing, using the ISO 14117 standard as a reference.
Methods
The research reported here did not involve animal or human experimentation and thus did not require institutional approval or need to follow ethical guidelines associated with the use of live vertebrate animals or humans.
Bioimpedance simulations
Simulation settings
Computable human models were simulated in Sim4Life (Zurich Med Tech AG, Zurich, Switzerland). First, the can of the simulated CIED was placed in the left pectoral region of the human anatomic models. Implantable cardioverter-defibrillator (ICD) and permanent pacemaker (PPM) leads then were placed into the heart through the cephalic vein with the cathode electrode positioned in the right ventricular (RV) apex while the opposite end of the lead was connected to the ICD/PPM connector (Figure 1D ). For ICD simulations, the distal shock coil was placed in the RV and the proximal shock coil was located in the superior vena cava.
Figure 1Computer-aided models for electrical safety implantable cardioverter-defibrillator (ICD) and permanent pacemaker evaluation during finite element model bioimpedance simulations. A: Male model hand-to-foot segmental body composition simulation with a smart scale. B: Male model wrist-to-finger segmental body composition simulation with a smart watch. C: Female model finger electrodermal activity simulation with a smart ring. D: Detail of the ICD simulated implant. The ICD lead is placed into the heart via the central veins with the proximal shock coil positioned in the superior vena cava, whereas the distal shock coil and pacing electrodes are positioned into the right ventricle.
To evaluate the voltage at the pacing electrodes for unipolar and bipolar pacing, all models were first discretized into small voxels to equally preserve the geometric properties of tissues and objects. The total number of voxels varied between simulations models. The lowest was 76.5 million voxels for unipolar PPM simulations, and the highest was 128.5 million voxels for bipolar PPM simulations. Simulations were performed with a workstation consisting of 2 processors (Xeon Gold 6226R, Intel, Santa Clara, CA), a graphic card (Quadro RTX 6000, NVIDIA, Santa Clara, CA), and 256 GB of memory.
We chose frequencies representative of use cases of commercial devices with bioimpedance sensing technology. Smart scale hand-to-foot simulations were performed by applying a sinusoidal current wave of 200 microamperes root mean square (μArms) with frequency of 3, 4, 5, 6, 7, 8, 9, 10, 50, 150 and 1000 kHz. Wearable smart watch wrist-to-finger simulations were performed at 5, 15, 20, 100, and 200 kHz applying a sinusoidal voltage wave of amplitude 1 volt peak to peak (Vpp). Single-frequency smart ring finger bioimpedance simulations were performed at 10 Hz with a sinusoidal voltage signal having amplitude of 1 Vpp using the inner electrodes of the ring to model the electrodermal activity (EDA) measurement.
Altogether, we ran 224 simulations for a total simulation time of 3200 hours, which included 176 smart scale simulations (11 frequencies × 2 left/right side × 4 configurations × 2 male/female model), 40 smart watch simulations (5 frequencies × 4 configurations × 2 male/female model), and 8 smart ring simulations (1 frequency × 4 configurations × 2 male/female model).
Simulated anatomic human models
We used as computable human phantoms the Virtual Family whole-body anatomic male models Fats version 3 (male, age 37 years, height 1.82 m, weight 120 kg, body mass index 36.2 k/m2) and female morphed Ella version 3 (female, age 26 years, height 1.63 m, weight 79.7 kg, body mass index 30 kg/m2).
These models provide representative values of individuals with overweight or obesity at risk for heart disease and originally were developed for electromagnetic and medical device safety evaluations. These models include more than 120 anatomic features and more than 300 tissues altogether. Electrical properties of the anatomic tissues and fluids were taken from the IT’IS low-frequency 4.0 database (IT’IS Foundation, Zurich, Switzerland) at the range of frequencies simulated.
For hand-to-foot smart scale simulations (Figure 1A), we posed the models and aligned the heels in the feet electrodes with the center of the electrodes evenly distributed on the surface of the heels while the fingers were positioned flat on the hand electrodes. Hand and foot current electrodes were modeled with surface areas of 31.9 and 106.7 cm2, respectively. For wrist-to-finger smart watch simulations, we posed the models wearing the B-Secur (Belfast, United Kingdom) HeartKey Test Watch below the left wrist bone, a watch developed to acquire simultaneous electrocardiographic and BIA measurements. The current source electrode (1.56 cm2) was positioned in the caseback of the watch in contact with the dorsum of the wrist while the current sink electrode (1.69 cm2) was positioned on the case to allow index-finger contact (Figure 1B). We simulated an EDA measurement placing a smart ring on the left fourth digit. The smart ring inner current electrodes had dimensions of 7 × 7 mm with an edge-to-edge distance of 4 mm (Figure 1C).
Simulated ICD, permanent cardiac pacemaker, and leads
Supplemental Table 1 provides a comparison of the mechanical dimensions of commercially available PPMs and ICDs vs our simulations (Figure 1D). For the simulations, we assigned the electrical material property of an ideal conductor for the metal outer shell of the can and polyurethane for the connector.
We designed a dual-coil ICD lead based on Reliant 4-FRONT lead specifications (Boston Scientific, Marlborough, MA) (Figure 1D). The dimensions are length 62.7 and 61.4 cm for Fats and morphed Ella models, respectively. On both models, the ICD lead diameter was 2.6 mm with annular cathode electrode length of 3 mm. The spacing between pacing electrodes was 1 cm with the anode ring electrode length of 3 mm. The spacing between the cathode and the distal shock electrode was 1 cm. The length of the distal and proximal shock coils was 5 and 8 cm, respectively, 0.15 mm in diameter, and spaced 10 cm apart.
We designed 2 different leads for bipolar PPM simulations based on the minimum and maximum spacing between electrodes from commercially available leads listed in Supplemental Table 2. The leads had the same length as the ICD lead for Fats and morphed Ella models, and the diameter was 2 mm. Pacing electrodes and can were assigned as ideal metal conductors while polyurethane was assigned to other nonmetal parts. The contact impedance between tissue and electrodes including their frequency behavior and active components (ie, functioning of CIED) cannot be simulated with the finite element method.
Benchtop bioimpedance testing
To evaluate a bioimpedance hardware system equivalent to our simulation study, we connected an impedance analyzer MFIA (Zurich Instruments, Zurich, Switzerland) to the input C terminal of the tissue-equivalent interface circuit defined by the ISO 14117 standard (Figure 2). Unlike smart scales or smart watch devices, this instrument gave us total control to change the frequency, the amplitude of the signal, as well as the signal measurement duration at each frequency. For each device tested from Boston Scientific, Medtronic (Minneapolis, MN), and Abbott (Chicago, IL), we performed a frequency sweep, increasing the frequency from 1 Hz to 1000 kHz. In order to save time, the frequency lists were chosen with more frequencies in the lower kilohertz range and fewer above 100 kHz, because the low-frequency region is the most restricted in the norm. At each of the 43 frequencies tested, we then performed a signal voltage amplitude sweep, increasing the voltage in steps from 2 millivolts peak-to-peak (mVpp) up to 2 Vpp (maximum value generated by the bioimpedance instrument used) or until interference was detected. For each frequency and amplitude tested, the total measurement duration was 60 seconds.
Figure 2Tissue-equivalent interface circuit for benchtop testing. A: Electrical connections specified in the ISO 14117 standard to check for electrical interference during benchtop testing connected to a common-mode, single-channel bipolar device under test. In differential mode, the single-channel bipolar device under test is connected between the coupled outputs F and G and the output J of the tissue-equivalent interface. B: Three-dimensional rendering of the tissue-interface built for benchtop testing to check for device overpacing or oversensing.
All devices were tested in basic pacing modes with standard programming, at highest sensitivity settings to simulate worst-case scenarios. Hardware used was that available to the investigators from clinical explants and from expired inventory; no Medtronic defibrillator lead was available for testing. No generator battery was at elective replacement interval. Only the RV lead was attached to the generator during testing, and atrial sensing was deactivated on all generators. Testing was performed during programmer connection, to assess for device detection of noise, and appropriate or inappropriate categorization of signals as noise vs inappropriate oversensing of noise signals. Sensing and pacing responses were noted. The specific tested hardware and setting configurations are given in Table 1. Abbott and Boston Scientific were tested in common mode (ie, creating the potential difference referenced to the local common or ground), with the anode and cathode connected to terminals F and G of the equivalent circuit, and the generator grounded through terminal J. Medtronic generator was tested in differential mode, that is, creating the potential difference between the 2 terminals anode and cathode connected between the coupled outputs F and G and the output J of the tissue-equivalent interface.
Table 1Characteristics and programming of tested devices
Segmental body composition simulations using a smart scale
Table 2 summarizes the voltage induced at the pacing electrodes for PPM and ICD during electrical current-driven finite element method (FEM) bioimpedance simulations across the legs, the arms, and the trunk. At frequencies above 3 and below 167 kHz, the ISO 14117 standard establishes a maximum voltage amplitude value that increases linearly by 6 mVpp/kHz. Above 167 and below 1000 kHz, the maximum voltage amplitude is set constant to 1 Vpp. As expected, due to the electrical properties of tissues and fluids at the range of frequencies simulated, the level of electrical interference varied with the frequency. The greatest variability was observed in the male model for both PPM and ICD simulations. Table 2 lists sex differences in the level of interference between models, with voltage values that were lower in the female model compared to the male model. The differences between models is due to both body dimensions and the pose of the phantoms because tissue properties were taken from the same database. There also was variation in the level of interference during left or right body composition simulation, with a higher level of interference in simulations on the left side of the body where the device is implanted. We found different levels of interference during PPM simulations in unipolar and bipolar configurations, with higher interference values at low frequency in unipolar mode due to the greater distance between anode and cathode. Finally, PPM results (Table 2) show that the level of electrical interference was below the threshold values determined by the ISO 14117 standard at frequencies starting at 3 and 10 kHz for female and male models, respectively. For ICDs, these frequencies are 5 and 50 kHz for female and male models, respectively. Considering the worst case at each frequency simulated, the safe current that does not induce an interference exceeding the threshold value of the norm increased with the frequency, with the most restrictive maximum safe current value being 51.2 μArms at 3 kHz.
Table 2Voltage induced at the PPM and ICD pacing electrodes during hand-to-foot segmental body composition bioimpedance simulations with a smart scale
Frequency
Male model
ISO 14117 threshold voltage
Safe current
PPM
ICD
Unipolar
Bipolar 15 mm
Bipolar 1.1 mm
Bipolar 10 mm
Left side
Right side
Left side
Right side
Left side
Right side
Left side
Right side
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
3
69.4
70.3
65.1
65.1
60.9
60.8
56.0
56.0
59.3
59.3
59.4
59.4
64.2
64.1
66.1
66.1
18
51.2
4
59.4
60.2
63.8
63.8
52.6
52.5
59.0
59.1
58.4
58.4
66.3
66.4
63.4
63.3
66.3
66.3
24
72.3
5
64.0
64.9
66.3
66.3
46.5
46.4
59.9
60.0
66.5
66.5
60.7
60.7
63.0
62.9
62.1
62.2
30
90.2
6
63.6
64.4
61.3
61.2
64.0
63.9
67.4
67.5
63.1
63.1
61.6
61.6
60.9
60.8
65.6
65.7
36
106.7
7
65.2
66.0
65.3
65.3
75.6
75.5
60.2
60.3
67.0
67.0
61.0
61.0
63.5
63.4
67.5
67.5
42
111.1
8
68.0
68.9
69.8
69.8
63.4
63.2
54.1
54.2
62.4
62.4
64.8
64.8
55.7
55.6
61.2
61.3
48
137.5
9
61.6
62.4
67.1
67.1
54.7
54.6
54.3
54.4
71.2
71.2
65.4
65.4
61.4
61.3
65.8
65.8
54
151.6
10
63.6
64.5
67.9
67.9
55.5
55.4
54.1
54.1
65.8
65.8
65.2
65.2
63.0
62.9
67.3
67.3
60
176.6
50
63.6
64.5
67.7
67.7
52.9
52.8
58.6
58.6
62.9
62.9
64.3
64.3
65.0
64.9
64.9
64.9
300
886.3
150
65.3
66.1
67.6
67.6
61.1
61.0
63.0
63.0
62.9
62.8
61.6
61.6
67.0
66.9
63.5
63.5
900
2662.1
1000
56.4
57.2
57.6
57.5
66.9
66.8
51.9
52.0
48.4
48.3
57.2
57.2
51.3
51.2
57.5
57.5
1000
2990.6
Frequency
Female model
ISO 14117 threshold voltage
Safe current
PPM
ICD
Unipolar
Bipolar 15 mm
Bipolar 1.1 mm
Bipolar 10 mm
Left side
Right side
Left side
Right side
Left side
Right side
Left side
Right side
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
3
36.2
41.1
22.0
23.2
28.0
28.0
7.0
7.4
34.6
34.6
17.5
17.6
35.8
35.8
25.5
25.8
18
51.2
4
37.1
42.0
17.1
18.3
32.8
32.8
7.0
7.4
35.0
35.0
20.2
20.2
38.7
38.7
25.9
26.3
24
72.3
5
39.8
44.6
24.1
25.3
28.5
28.5
6.9
7.3
36.6
36.6
18.0
18.1
29.9
29.9
24.0
24.3
30
90.2
6
36.5
41.4
22.6
23.8
29.2
29.2
7.0
7.4
26.2
26.2
8.6
8.6
28.4
28.4
17.1
17.4
36
106.7
7
38.8
43.7
23.9
25.1
24.6
24.6
7.0
7.4
35.0
35.0
8.6
8.7
28.7
28.7
26.3
26.7
42
111.1
8
35.4
40.3
17.4
18.6
31.9
31.9
7.0
7.4
30.5
30.5
11.1
11.2
37.9
37.9
24.9
25.2
48
137.5
9
38.5
43.4
24.6
25.8
27.4
27.4
7.0
7.4
33.9
33.9
24.3
24.3
38.2
38.2
14.5
14.8
54
151.6
10
36.0
40.9
13.1
14.2
29.7
29.7
7.0
7.4
34.4
34.4
19.8
19.9
37.1
37.1
16.1
16.5
60
176.6
50
34.2
39.1
27.8
29.0
26.6
26.6
7.0
7.4
36.9
36.9
21.9
21.9
35.6
35.6
27.1
27.4
300
886.3
150
35.5
40.4
26.3
27.5
35.6
35.6
7.0
7.4
40.0
40.0
17.2
17.3
35.6
35.6
28.9
29.3
900
2662.1
1000
34.0
38.9
22.3
23.5
21.0
21.0
6.9
7.3
36.1
36.1
24.9
24.9
34.0
34.0
20.5
20.8
1000
2990.6
The results are the induced voltage in millivolts peak-to-peak (mVpp) recorded at the anode and the cathode in unipolar and bipolar configurations measuring left- and right-body composition bioimpedance at frequencies of electrical current from 3 to 1000 kHz with 200 microamperes root-mean-square (μArms). The safe current column contains the amplitude of current (in μArms) that is safe to inject to not exceed the maximum ISO 14117 threshold voltage. The highlighted cells correspond to values that exceed the maximum allowed by the ISO 14117 standard.
Segmental body composition simulations using a smart watch
Table 3 summarizes the voltage induced at the pacing electrodes for PPM and ICD during left wrist to right thumb and right index finger voltage-driven FEM bioimpedance simulations. Induced voltage values are below the ISO 14117 threshold starting at 100 kHz in both male and female models. Unlike hand-to-foot current driven simulations (Table 2), the induced voltages during wrist-to-finger simulations in the female model are higher than in the male model. Interference voltages produced during voltage-driven simulations (Table 3) are larger than those reported with current-driven simulations (Table 2). Considering 5 kHz as the worst case, the maximum safe voltage value that does not induce an interference exceeding the threshold value of the standard is 66 mVpp.
Table 3Voltage induced at the PPM and ICD pacing electrodes during wrist-to-finger segmental body composition bioimpedance simulations with a smart watch
Frequency
Male model
ISO 14117 threshold voltage
Safe voltage
PPM
ICD
Unipolar
Bipolar 15 mm
Bipolar 1.1 mm
Bipolar 10 mm
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
5
390.9
391.7
390.2
390.1
391.2
391.2
390.8
390.7
30
66.4
15
392.0
392.9
389.9
389.8
391.6
391.6
390.6
390.5
90
199.4
20
391.5
392.4
391.0
390.9
391.3
391.3
394.8
394.7
120
265.4
100
390.2
391.1
391.3
391.2
390.6
390.5
389.7
389.6
600
1334.9
200
387.1
388.0
386.7
386.6
387.3
387.3
387.3
387.2
1000
2237.8
Frequency
Female model
ISO 14117 threshold voltage
Safe voltage
PPM
ICD
Unipolar
Bipolar 15 mm
Bipolar 1.1 mm
Bipolar 10 mm
Anode
Cathode
Anode
Cathode
Anode
Cathode
Anode
Cathode
5
447.3
447.9
451.3
451.3
451.8
451.8
449.0
449.0
30
66.4
15
445.4
446.1
451.4
451.3
449.1
449.1
450.3
450.2
90
199.4
20
446.8
447.4
451.7
451.7
452.1
452.1
450.2
450.2
120
265.4
100
448.8
449.5
447.5
447.5
449.1
449.1
448.8
448.7
600
1334.9
200
445.3
446.0
446.9
446.8
443.5
443.5
445.0
445.0
1000
2237.8
The results are the induced voltage in millivolts peak-to-peak (mVpp) recorded at the anode and the cathode in unipolar and bipolar configurations measuring wrist-to-finger segmental body composition bioimpedance at frequencies of electrical current from 5 to 200 kHz with 1 volt peak to peak (Vpp). The safe current column contains the amplitude of voltage (in mVpp) that is safe to apply to not exceed the maximum ISO 14117 threshold voltage. The highlighted cells correspond to values that exceed the maximum allowed by the ISO 14117 standard.
Table 4 summarizes the voltage induced at the pacing electrodes for PPM and ICD during single-frequency finger EDA voltage-driven FEM bioimpedance simulations. Although an EDA measurement using a smart ring is mostly focused around the finger, Table 4 shows the distribution of electric potential induced in the PPM and ICD pacing electrodes exceeds the 2 mVpp maximum value allowed by the ISO 14117 standard. The level of interference is higher in the female model than in the male model, with higher voltage values in bipolar configuration than in unipolar configuration.
Table 4Voltage induced at the PPM and ICD pacing electrodes during finger electrodermal simulation with a smart ring
Extrapolation from 16.6 Hz value specified in the ISO 14117 standard.
The results are the induced voltage in millivolts peak-to-peak (mVpp) recorded at the anode and the cathode in unipolar and bipolar configuration measuring electrodermal activity at frequency of electrical current 10 Hz with 1 volt peak to peak (Vpp). The safe current column contains the amplitude of voltage (in mVpp) that is safe to apply to not exceed the maximum ISO 14117 threshold voltage. The highlighted cells correspond to values that exceed the maximum allowed by the ISO 14117 standard.
Considering the worst case simulated (female with PPM in bipolar configuration), the maximum voltage amplitude that does not exceed the threshold value of the norm at 10 Hz is 20 mVpp.
Benchtop results
Benchtop interference test results are summarized in Table 5. For each tested frequency from 1 Hz to 1000 kHz, the minimum bioimpedance voltage amplitude at which we observed the effect of electrical interference in the real-time electrocardiograms. Abbott and Boston Scientific were tested in common mode only because this mode setting already interfered with the devices. Medtronic cardiac resynchronization therapy–defibrillator (CRT-D) is the only device that was tested in both common and differential modes. Unlike Abbott and Boston Scientific CRT-Ds, no oversensing or overpacing was observed with the Medtronic CRT-D in common mode when measuring bioimpedance over the entire range of frequencies and amplitudes tested. We only observed interference to Medtronic CRT-D in differential mode, and these are the results reported here. The range of interference varied with signal amplitude and frequency between 5 and 40 Hz for the Boston Scientific generator; between 10 and 300 Hz, and 500 Hz and 1 to 2 kHz for the Abbott generator; and from 1 to 2 kHz for the Medtronic generator.
Table 5Data on the outcome of CRT-D during bioimpedance benchtop testing at various frequencies and voltage amplitude levels
Frequency (Hz)
Bioimpedance signal amplitude (mV peak)
Boston Scientific
Abbott
Medtronic
1
—
—
200
2
—
—
50
3
—
—
5
4
—
—
5
5
1000
—
5
6
500
—
1
7
1000
—
1
8
1000
—
1
9
1000
—
1
10
1000
1
1
20
700
400
1
30
600
300
1
40
600
200
1
50
—
200
1
60
—
200
1
70
—
200
1
80
—
200
1
90
—
200
1
100
—
200
1
200
—
300
10
300
—
600
20
400
—
—
10
500
—
800
15
600
—
—
15
700
—
—
15
800
—
—
15
900
—
—
25
1000
—
500
35
2000
—
1
1000
3000
—
—
—
4000
—
—
—
5000
—
—
—
6000
—
—
—
7000
—
—
—
8000
—
—
—
9000
—
—
—
10,000
—
—
—
15,000
—
—
—
20,000
—
—
—
50,000
—
—
—
150,000
—
—
—
200,000
—
—
—
1,000,000
—
—
—
For each frequency evaluated (in Hertz), we determined the minimum voltage amplitude (in millivolts peak) that caused electrical interference in the form of oversensing or overpacing.
Across device manufacturers, CRT-D generators demonstrated susceptibility to oversensing and pacing inhibition at signal amplitude and frequencies tested. Representative tracings for each manufacturer at selected bioimpedance frequencies and signal voltage levels are shown in Figure 3. Figure 3A shows oversensing then appropriate “noise” categorization on an Abbott CRT-D transitioning from 0.8 to 1 Vpp at 50 Hz. Figure 3B shows oversensing and pacing inhibition on an Abbott CRT-D at 110 mVpp and 1 kHz. Figure 3C shows noise without detection on a Boston Scientific CRT-D at 2 Vpp and 5 Hz. Figure 3D shows intermittent oversensing and noise categorization on a Boston Scientific CRT-D at 1.4 Vpp and 200 Hz. Figure 3E shows oversensing on a Medtronic CRT-D at 10 mVpp and 200 Hz. Figure 3F shows noise without detection on a Medtronic CRT-D at 50 mVpp and 900 Hz.
Figure 3Sample cardiac implantable electronic device tracings across vendors. A: Shows oversensing then appropriate “noise” categorization on an Abbott CRT-D. B: Shows oversensing and pacing inhibition on an Abbott CRT-D. C: Shows noise without detection on a Boston Scientific CRT-D. D: Shows intermittent oversensing and noise categorization on a Boston Scientific CRT-D. E: Shows oversensing on a Medtronic CRT-D. F: Shows noise without detection on a Medtronic CRT-D. See the text for testing details.
We performed electrical interference evaluations of bioimpedance technology available on smart scales, smart watches, and smart rings on CIEDs following the FDA’s accepted ISO 14177 standard. First, we evaluated the level of electrical interference on a male and a female computable human model with simulations. Our approach facilitates consistency, predictability, and reproducibility of the simulation results because they are not specific to a particular cohort, nor are they affected by experimental confounding factors such as additional electromagnetic interference from external sources. Another advantage of simulations is the ability to measure and visualize the voltage and current flowing through any part of the human model (Supplemental Figure 1). A challenge associated with realistic simulations such as ours is verifying that the results are accurate, given the high number of simulation parameters. For this, we first set our simulation settings modeling a basic geometry for which there are theoretical bioimpedance models available. The agreement between predicted theoretical and simulation voltage values was excellent, with a relative error <0.05%. These analyses were then extended to determine the level of interference generated with bioimpedance following ISO 14117 test methodologies using a tissue-interface equivalent circuit. The test bench study adopted here allows technical requirements to be standardized in order to be able to perform a head-to-head comparison of interference in CIEDs from different manufacturers under a variety of reproducible test cases and conditions.
Our results are novel and timely because consumer devices with bioimpedance sensing capabilities already are commercially available. These results expand the current knowledge on electromagnetic interference with CIEDs.
Electromagnetic interference in cardiac electronic implants caused by novel electrical appliances emitting electromagnetic fields in the intermediate frequency range: a systematic review.
Static magnetic field measurements of smart phones and watches and applicability to triggering magnet modes in implantable pacemakers and implantable cardioverter-defibrillators.
Simulations considering smart scale, wearable smart watch, and smart ring use cases show the level of bioimpedance interference depends on the signal frequency, signal amplitude, sex, body segment measured, distance between anode and cathode, pacing mode, and strategy used to measure bioimpedance. Additional experimental factors that we found to influence the extent of bioimpedance electrical interference in CIEDs during benchtop testing included gain and sensitivity settings.
The only existing studies that have evaluated electrical interference during bioimpedance measurements in CIEDs focused on monitoring device anomalies and adverse events in patients with these devices.
reported no anomalies in 20 patients with ICDs undergoing bioimpedance measurements using the InBody 520 instrument (Seoul, South Korea). Subjects were measured across both legs, arms, and the trunk at 5, 50, and 500 kHz, but the signal amplitude and the duration of exposure were not provided. Similar results were reported by Meyer et al
for 63 patients equipped with ICDs using the Nutriguard MS instrument (DataInput Gmbh, Darmstadt, Germany) measuring hand to foot at 5, 50, and 100 kHz with a current amplitude of 800 μA. The results of Meyer et al were consistent with those from recent studies by Chabin et al
using the same instrument. In these 2 studies, the authors analyzed separately the results obtained by different CIED manufacturers but did not assess the impact of duration of exposure or lead characteristics.
Electromagnetic interference in cardiac electronic implants caused by novel electrical appliances emitting electromagnetic fields in the intermediate frequency range: a systematic review.
The authors found no changes in pacing thresholds nor did they observe inappropriate sensing in intracardiac electrograms measuring over 200 patients, regardless of the body side (left or right) measured,
did report the occurrence of arrhythmic events on the same day of the measurement in 6 patients with a recent history of documented paroxysmal atrial fibrillation (1 subject) and recurrent ventricular tachycardia (5 subjects), but the authors determined these to be coincidental and not triggered by bioimpedance testing.
Despite these observational studies asserting that bioimpedance can be safely performed in patients equipped with CIEDs,
our results suggest that analyzing only disturbances in CIED behavior attributable to electrical interference might be insufficient to corroborate bioimpedance safety in patients with these devices. Reported simulation data show evidence of potential interference based on FDA-recognized threshold voltages at the pacing electrodes defined in the ISO 14117 standard. Case in point, considering 5 kHz as the lowest and most restrictive frequency measured by Meyer et al
our simulations suggest the maximum safe current is 90.2 μArms, which is 3 times less current than that applied in these studies. Also, our benchtop results suggest that, at the tested voltages and frequencies, commercially available devices with bioimpedance technology could exceed current standards and at least have the theoretical “potential” to interfere, with undesirably disastrous consequences. Oversensing or overdetecting could lead to (1) withholding lifesaving pacing in a pacer-dependent patient or (2) inappropriate shocks in an ICD patient, often resulting in posttraumatic stress disorder.
Although manufacturers use filters in their CIEDs to attenuate signals out of the physiological frequency range of the heart, detailed information on these protection circuits as well as noise detection algorithms are proprietary and differ among manufacturers (example given in Table 5). The Medtronic generator was the only one of the devices tested to have no interference in common mode, thus suggesting that both internal circuitry and algorithm are more robust against common-mode bioimpedance interference than the Abbott and Boston Scientific generators tested.
Finally, the technical specifications of consumer devices with bioimpedance technology vary according to manufacturer, device, and intended use case, thereby requiring a case-by-case objective and quantitative interference evaluation. Given our findings, it is imperative to extend this study to test broader variety of hardware and settings and then test patients with CIEDs to determine the translation of our findings to patients using such consumer electronics.
Study limitations
First, the ISO 14117 standard does not define specific tests for devices with bioimpedance technology. Second, the voltage threshold values at the pacing electrodes are defined in the ISO 14117 standard via single-ended bench voltage measurements, for which the CIED can is physically connected to a 0 V reference voltage point or electrical ground (terminal J in Figure 2A) from which these voltages are measured. However, in real life, patients with or without CIEDs who undergo bioimpedance measurements are intentionally not grounded to protect them from potentially harmful electrical shock hazards and leakage currents as defined by the standard International Electrotechnical Commission 60601 for medical electrical equipment.
International Electrotechnical Commission, IEC 60601-1, 2015, Medical electrical equipment. https://webstore.iec.ch/publication/2603. Accessed December 21, 2022.
Third, the results obtained from testing the tissue interface defined by the ISO 14117 standard are limited because the standard assumes an equivalent electrical circuit that represents a simplification of the body impedance. Another limitation of the tissue interface is that it does not include the effect of tissue-electrode contact impedances, which will impact the correct functioning of CIEDs during bioimpedance testing. Fourth, although computable human models are more realistic than a tissue-equivalent circuit, they are still approximations of human anatomy and therefore have limitations. Fifth, benchtop testing was performed with CRT-Ds programmed at the highest sensitivity in bipolar configurations, to emulate a “worst-case scenario.” More conservative programming and/or unipolar configurations, also including a conduction medium, may generate different results. Testing was performed on hardware available to the investigators due to clinical explant (generators) and/or expired stock (leads). Therefore, limited combinations of generators/leads were available for testing. We then focused on RV leads only because those were the leads available and they would be most consequential for interference—oversensing leads to both inhibition of pacing and inappropriate detection of ventricular tachycardia/ventricular fibrillation. Lastly, the translation of these in silico findings and benchtop testing using direct connections to CIED leads, to sensed signals including the effect of electrode contact impedance across the human body, remains unknown and awaits further investigation in future (pre-)clinical studies.
Conclusion
This study provides benchmark data to evaluate the safety of smart scales, smart watches, and smart rings with bioimpedance technology on CIEDs. The methodology and results presented represent the most comprehensive technical analysis today available on the level of interference induced by bioimpedance devices on CIEDs using the FDA-accepted ISO 14117 standard as a reference. Our results suggest that consumer electronic devices with bioimpedance technology could induce an electric voltage exceeding ISO 14117 maximum values for CIEDs. These serious adverse events may occur without forewarning in patients with CIEDs and have the potential to interfere with lifesaving therapy when these patients undergo bioimpedance measurements using smart scales, wearable smart watches, and smart rings. The present findings do not recommend the use of these devices with bioimpedance technology in this population due to potential electrical interference. These results also call for a review and update of the ISO 14117 standard to define new specific tests for devices with bioimpedance technology. Future clinical studies should take our findings into account and include further confirmatory simulation and benchtop studies considering worst-case conditions to determine clinical relevance and corroborate the safety of smart scales, wearable smart watches, and smart rings with bioimpedance technology in patients with CIEDs.
Acknowledgments
The authors would like to thank Dr Rob S. Macleod from the Scientific Computing and Imaging Institute, Nora Eccles Cardiovascular Research and Training Institute, and Biomedical Engineering Department at the University of Utah for his comments on an early version of this work. Files to build the tissue-interface printed circuit board and the schematic for electronic circuit are available upon request to the corresponding author.
Analysis of different device-based intrathoracic impedance vectors for detection of heart failure events (from the Detect Fluid Early from Intrathoracic Impedance Monitoring Study).
ANSI/AAMI/ISO 14117, Active implantable medical devices—Electromagnetic compatibility—EMC test protocols for implantable cardiac pacemakers, implantable cardioverter defibrillators and cardiac resynchronization devices, 2019. https://www.iso.org/standard/73915.html. Accessed December 21, 2022.
Electromagnetic interference in cardiac electronic implants caused by novel electrical appliances emitting electromagnetic fields in the intermediate frequency range: a systematic review.
Static magnetic field measurements of smart phones and watches and applicability to triggering magnet modes in implantable pacemakers and implantable cardioverter-defibrillators.
International Electrotechnical Commission, IEC 60601-1, 2015, Medical electrical equipment. https://webstore.iec.ch/publication/2603. Accessed December 21, 2022.
Funding Sources: Research reported in this publication was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number K23HL143156 to Dr Steinberg. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Disclosures: Dr Sanchez holds equity in Haystack Diagnostics, Inc., which has an option to license patented bioimpedance technology for neuromuscular evaluation where the author is named an inventor; holds equity and serves as Scientific Advisory Board Member of Ioniq Sciences, Inc., which commercializes bioimpedance-based technology for early cancer detection; holds equity and serves as Scientific Advisor to the Board of B-Secur, Ltd, which commercializes electrocardiography and bioimpedance technology; consults for Myolex, Inc., which has an option to license patented bioimpedance technology where the author is named an inventor; serves as a consultant to ImpediMed, Inc., which commercializes bioimpedance technology for fluid assessment and has patented bioimpedance technology where the author is named an inventor; and serves as a consultant in bioimpedance applications to Texas Instruments, Inc., Happy Health, Inc., Analog Devices, Inc., and Eko Health, Inc. Dr Steinberg reports research support from AHA/PCORI, Abbott, Cardiva, Sanofi, and AltaThera; and consulting to Sanofi, InCarda, Milestone, Pfizer, and AltaThera. All other authors have no conflicts on interest to disclose.
02696-0&id=fx1.jpg){kind=link}
02696-0&id=gr1.jpg){kind=link}
02696-0&id=gr2.jpg){kind=link}
02696-0&id=gr3.jpg){kind=link}