Comparative Accuracy of a Wireless Heart Rate Monitor (NeoBeat) Versus Standard EKG During Resuscitation of Very Premature Neonates

Abstract

Background:

Rapid and accurate heart rate assessment is essential in neonatal resuscitation, particularly for very premature infants where delays can impact timely intervention. A wireless heart rate monitor (NeoBeat) device has been validated in term babies but not in smaller preterm infants. We sought to validate the use of the Neobeat device compared to standard EKG in extremely preterm neonates during resuscitation.

Methods:

A prospective observational study of very premature neonates undergoing routine resuscitation was conducted at a tertiary care center. Heart rate data were simultaneously recorded using the NeoBeat wireless monitor and a standard three-lead EKG.

Results:

Thirty very preterm infants (gestational age, (GA) 26.6 ± 1.9 weeks and birth weight range (250–1310 g)) were enrolled. A total of 1690 heart rate pairs were analyzed. A strong correlation and tracking observed between NeoBeat and EKG-derived heart rates (Pearson r = 0.96, Spearman ρ=0.95), with a mean absolute difference of 3.76 beats per minute (bpm). Correlation remained robust across different resuscitation interventions (Intubation: r = 0.93 vs. No Intubation: r=0.90, p =0.207, PPV: r=0.93 vs. No PPV: r = 0.87, p=0.726).

Conclusion:

In very preterm infants, NeoBeat wireless monitor achieved 96% precision (r=0.96) and maintained a mean absolute difference of 3.76 bpm compared to EKG, supporting its reliability and accuracy for real time neonatal monitoring, including infants weighing as little as 250 grams, though statistical power in this range was limited. Randomized studies comparing NeoBeat to standard EKG are warranted to confirm these findings in extremely preterm infants and explore whether NeoBeat improves clinical outcomes.

Introduction

Timely and accurate heart rate assessment is central to guideline-based neonatal resuscitation. Delays in heart rate assessment can impede the initiation of essential steps based on the neonatal resuscitation algorithms such as the Neonatal Resuscitation Program (NRP) or Newborn Life Support (NLS). This is even more critical for very premature infants who need meticulous and timely cardiopulmonary support or may develop hypoxia or hypoperfusion which may be associated with serious morbidities such as intraventricular hemorrhage or death.¹ Placement of traditional three-lead electrocardiography (EKG), in addition to pulse oximetry, takes several minutes to display a stable heart rate², which may alter neonatal interventions at birth.¹ In addition, placement of these monitors ties up resuscitation personnel, which may be challenging in resource-limited scenarios.³

Wireless, rapidly deployable technologies such as the NeoBeat device (Laerdal, Switzerland) offer a potential solution to these logistical and temporal barriers, with studies suggesting faster acquisition of reliable heart rate values compared to EKG.^4–7 Because these devices do not require adhesives, they can be placed on infants within a few seconds, allowing the provider to focus on other aspects of neonatal care.^4,6

Various methods have been tested during neonatal resuscitation to assess heart rate; 3 lead EKG however is still a preferred method due to its speed and accuracy over pulse oximetry.^8–10 Other methods such as auscultation and cord palpation are possible in settings where EKG is unavailable but have associated limitations in precision and varying accuracies.^{8 11} Furthermore, data from low-resource settings show that easily applied wireless monitors can address infrastructure gaps by obviating the need for bedside monitors, cables, or electrodes.^3,7 To strengthen the clinical evidence-base for this emerging technology, we evaluated the correlation and agreement between NeoBeat derived heart rates and those obtained via conventional three-lead EKG in a cohort of very to extremely premature neonates.

Methods

This prospective observational study was conducted at a tertiary care center where routine monitoring at delivery is standard practice for premature or at-risk infants. The study utilized an IRB approved delivery room database to collect deidentified demographic information which met the requirements for a waiver of consent. An Advanced Neonatal Life Support team—comprised of a Neonatologist or Neonatal Nurse Practitioner, a designated Registered Nurse, and a Respiratory Therapist—attended each qualifying delivery. Overhead-mounted cameras recorded neonatal resuscitations for quality assurance and training.⁷ All infants included in this analysis were between 23 and 30 weeks of gestation at birth. The NeoBeat device was placed on infants during delayed cord clamping. Video review confirmed the precise timeframe during which both monitors (NeoBeat and traditional 3 lead EKG) were simultaneously active, and timestamps of resuscitation events were logged independently for all 30 subjects. Heart rate measurements on conventional EKG monitor (Carescape Model B450; GE Healthcare) are stored in real-time in the same computer as the video files and therefore share the same timestamps. As illustrated in Figure 1, equipment application was observed from the overhead video angle, allowing precise timing of device placement and signal appearance.

Figure 1.

Resuscitation scene from the delivery room video perspective. The illustration depicts the resuscitation team and equipment layout as viewed from the ceiling-mounted video camera. NeoBeat and EKG leads are shown being applied simultaneously, highlighting the real-time context in which placement and signal acquisition times were recorded. All events were logged independently during video review for objective timing analysis.

All data management, signal alignment, and statistical analyses were performed in Python 3.9.12 (Python Software foundation) using the following packages: pandas, NumPy, SciPy, sklearn, matplotlib. NeoBeat and EKG signals were manually aligned using a custom Python visualization tool that allowed precise synchronization without interpolation or filtering. All gaps and missing values, as was as physiologically improbable values were retained to preserve the real-world timing and integrity of both outputs.

Pearson correlation coefficient was calculated to gauge linear relationships, Spearman correlation coefficients were calculated to evaluate rank-order alignment and mean absolute difference (MAD) and root mean squared error (RMSE) to quantify measurement differences. Subgroup analyses were performed to examine whether gestational age, birth weight, or specific respiratory interventions (e.g., Continuous positive airway pressure (CPAP), intubation, or positive pressure ventilation (PPV)) significantly affected the NeoBeat EKG correlation. Statistical significance was set at p < 0.05.

Results

Thirty infants were included in the study (Figure 2). The median gestational age was 26.2 weeks [IQR: 25.0–27.9], and the mean birth weight was 880.6 ± 300.6 g (Table 1). The demographic characteristics of the cohort are summarized in Table 1, including length, fronto-occipital circumference (FOC), Apgar scores, sex, delivery mode, and respiratory interventions (CPAP, PPV, and intubation).

Figure 2.

Flow diagram of infants included in the study.

Table 1.

Demographics of infants with both NeoBeat and EKG monitoring.

Demographics Table (N = 30)
Variable	Value
Gestational Age (weeks)	26.6 ± 1.9 (Range: 23.3–30.6)
Birth Weight (g)	880.6 ± 300.6 (Range: 250–1310)
Length (cm)	34.3 ± 4.3 (Range: 22.8–40.0)
Fronto-Occipital Cir. (cm)	27.5 ± 7.1 (Range: 17.5–49.5)
Apgar Score at 1 min	4.5 [2.0–6.0]
Apgar Score at 5 min	7.0 [6.0–8.0]
Apgar Score at 10 min	6.0 [6.0–8.0]
Sex (Male)	17 (56.7%)
Delivery Mode (c/s)	28 (93.3%)
CPAP n (%)	28 (93.3%)
PPV n (%)	24 (80.0%)
Intubation n (%)	13 (43.3%)

G: grams, cm: centimeter, Cir: circumference, c/s: cesarean section, n: number of subjects.

The median time to place EKG leads from birth was 17.5 s [IQR 14.0–26.8], followed by a median of 50 s [IQR 27.0–72.0] to acquire a usable signal, for a total median delay of 73 s [IQR 48–88] from lead placement to signal detection. In contrast, NeoBeat placement required a median 4 s [IQR 2–4] and the first readable heart-rate value appeared 12 s [IQR 12–18] later (12 video-verified infants). For the remaining 18 infants, NeoBeat was applied off-camera while the umbilical cord remained intact during delayed-cord clamping; therefore, placement timing could not be extracted.

A total of 1,690 paired heart rate measurements were collected from 30 neonates at delivery. The MAD was 3.76 bpm, RMSE was 7.21 bpm, reflecting generally low error across the cohort, (Figure 3). The Pearson Correlation coefficient was r = 0.96, and the Spearman rank correlation was ρ = 0.95, indicating strong agreement in both magnitude and trend.

Figure 3.

A. Scatter plot showing paired heart rate measurements (n = 1,690) from NeoBeat and 3-lead EKG. The dashed identity line (y=x) indicates strong correlation (r=0.96) with minimal bias. B. Bland-Altman plot showing mean difference of −0.78 bpm, with 95% limits of agreement from approximately −14 to +14 bpm.

To assess correlation in the bradycardic range (Figure 4), we examined all paired heart rate values where EKG readings were under 100 bpm. Across 229 such observations from 18 infants, Pearson correlation remained strong (r = 0.799), with a mean difference of +2 bpm and limits of agreement from −15 to +20 bpm. A 30-second rolling average further improved correlation (r = 0.831), suggesting transient variability contributed to momentary divergence.

Figure 4.

(A) Scatter plot and (B) Bland-Altman plot comparing NeoBeat and EKG heart rates for paired values where EKG HR <100 bpm (n = 229). Correlation remained strong (r = 0.799), with a mean difference of +2 bpm (LoA: −16 to +20).

Subgroup analyses showed agreement across respiratory interventions (CPAP-only: r = 0.948; n = 6; PPV-only: r = 0.974; n = 2). When stratified by intubation status, correlation was similarly strong (intubated: r = 0.972; n = 13; not intubated: r = 0.950; n = 17). Notably, 22 of 30 infants (73%) received both CPAP and PPV, limiting interpretability of comparisons between individual support types due to overlapping exposure. Furthermore, ventilation mode annotations were not time-synchronized with heart rate data, precluding direct linkage of HR values to specific interventions

To assess signal reliability in the most vulnerable patients, we analyzed the first quartile of gestational age and birth weight. Both groups retained strong agreement (GA: r = 0.946; BW: r = 0.967), although the small subgroup size limits generalizability. These subgroup results are illustrated in Figure 5. There were no significant associations between NeoBeat–EKG correlation and gestational age (r = −0.07, p = 0.744), birth weight (r = −0.09, p = 0.680), body length (r = −0.15, p = 0.507), or fronto-occipital circumference (r = 0.18, p = 0.400).

Figure 5.

Scatter plot and Bland Altman for subgroups. Panel A and B represent the scatter plot and bland Altman plot respectively, for the lowest quartile gestational age (GA) subgroup. Panel C and D represent the scatter plot and bland Altman plots, respectively for the lowest quartile birth weights our cohort.

Discussion

This study demonstrates that the NeoBeat wireless heart rate monitor provides highly comparable measurements to conventional EKG in very premature infants undergoing resuscitation. Our findings showed minimal bias and narrow limits of agreement, consistent with prior studies demonstrating that wireless or dry-electrode technologies can capture neonatal heart rate quickly and achieve strong agreement with standard multi-lead EKG readings.^6,12 There are several advantages of a wireless heart rate monitor during neonatal resuscitation. The advantage of applying the device immediately after birth—often before infant transfer to the resuscitation area—reduces the time to first reliable heart rate reading instead of waiting until the baby reaches the warmer where wired monitors can be placed.^4,6 In our study the Neobeat could display a heart rate a full minute earlier than traditional EKG lead placement. This allows potential monitoring during intact cord resuscitation which can improve decision making of the neonatal team. As seen in the VentFirst trial where intact cord resuscitation was conducted on very preterm infants, heart rate detection was determined through cord palpation. This method is concerning for several reasons, first, palpation of the umbilical artery requires compression of the umbilical vein and in animal models this has been shown to cause a surge in carotid artery blood flow.¹³ Second, studies have shown that this consistently underestimates the true heart rate which may lead to unnecessary interventions such as positive pressure ventilation or early clamping of the umbilical cord.^14–16

Conversely, display of the heart rate on the infant’s chest may minimize delays in heart rate acquisition, decrease cognitive load on the resuscitation team, improve adherence to protocols, and potentially enhance neonatal outcomes. This enhanced workflow could be especially useful in crowded delivery rooms or resource-limited settings where conventional monitors are less readily available. Barriers to providing effective resuscitation for newborns include lack of proper equipment due to costs and accessibility.³ The ability to provide timely and effective resuscitation decreases the risk of intrapartum-related neonatal adverse outcomes.³ Detecting lower heart rates in the first minute of life has significant implications, as two-thirds of newborns requiring resuscitation have heart rates below 100 in the first 30 seconds of life,¹⁷ suggesting that any technology that speeds up heart rate measurement could have a meaningful impact on clinical decision-making. As wireless monitoring technologies evolve, their integration into standardized neonatal resuscitation protocols warrants further exploration, particularly in settings where accurate heart rate acquisition is essential, and reliable equipment may be unavailable.^3,16 The small sample size (n=30) limits the statistical power to detect subtle differences across gestational age strata (by weeks) or specific respiratory interventions. The EKG device averaged heart rate values over 5-second intervals whereas NeoBeat updated every second, introducing inherent differences in signal alignment. EKG waveforms were not captured to conduct a beat-by-beat or second-by-second comparison. Therefore, both EKG and NeoBeat samples were analyzed at 0.2 Hz by extracting the fifth-second value. Occasionally, large deviations were observed in one signal or the other, making it difficult to determine which value was physiologically accurate. None of the infants received chest compressions in the cohort even though there were brief tracings where the heart rate from the NeoBeat dropped below 60 beats per minute. It is likely the team was not responding to some of these readings which often occurred at the time of device removal or placement. Lastly, this was an observational study; a randomized design would be more robust in determining how faster heart rate acquisition might directly influence clinical outcomes.¹⁶

Although the device showed strong correlation in the entire cohort, anecdotal observations suggest that, among the smallest neonates with highly compliant chest walls, the NeoBeat’s physical structure may slightly impede spontaneous chest wall expansion. It is unclear whether restricting the chest wall may actually prevent overdistension of the lung. There were no reports of skin damage or adverse effects of the device on newborns. Future research should formally assess this concern—potentially by measuring tidal volumes during device application—and determine whether a lighter or smaller form factor could better accommodate very premature populations.¹⁸ To explore agreement in the smallest infants, we analyzed the first quartile of gestational age and birth weight. Both groups retained strong agreement (GA: r = 0.946; BW: r = 0.967), although the small subgroup size limits generalizability (figure 5).

The NeoBeat offers a rapid, accurate heart rate assessment in very premature neonates, aligning closely with standard three-lead EKG readings. By circumventing some of the logistical constraints associated with conventional monitors, NeoBeat may help neonatal teams adhere more consistently to newborn resuscitation guidelines and potentially improve early intervention effectiveness.¹⁶ As wireless technology evolves, optimizing device design for extremely low birth weight (ELBW) infants, integrating real-time data streams into established hospital workflows, and quantifying the impact on team cognitive load will be crucial.³ While prior trials have not shown any benefit from ECG vs no ECG monitors for clinical outcomes, their ability to provide an accurate heart rate and relieve a team member from auscultation may be an advantage.⁹ Further, large-scale investigations are needed using wireless ECG monitors and explore whether these devices can translate into improvements in resuscitation and neonatal outcomes.

Abbreviations

EKG

Electrocardiogram

GA

Gestational Age

BPM

Beats Per Minute

MAD

Mean Absolute Difference

RMSE

Root Mean Squared Error

PPV

Positive Pressure Ventilation

CPAP

Continuous Positive Airway Pressure

DTW

Dynamic Time Warping

FOC

Fronto-Occipital Circumference

ALS

Advanced Life Support

NRP

Neonatal Resuscitation

Program ELBW

Extremely Low Birth Weight

Data Statement:

The authors have agreed to share data on request by emailing the corresponding author anup.katheria@sharp.com