Introduction

The measurement of heart rate (HR) in beats per minute (bpm) is a fundamental physiological parameter derived from assessing the temporal duration between successive heart cycles initiated by the sinoatrial node1. HR serves as a crucial indicator for evaluating both general health and athletic performance2,3,4. Traditionally, HR is assessed using electrocardiography (ECG)5, utilizing either multiple-lead channels or basic chest straps equipped with two electrodes. However, this method is not ideal for prolonged monitoring during daily activities6. Photoplethysmography (PPG), especially when utilizing green light7, has gained increasing popularity in consumer wearables such as watches, earbuds, armbands, and rings. This technology, known for its reduced susceptibility to motion artifacts compared to infrared, significantly enhances the accuracy of optical HR monitors and has established PPG as the standard method for continuous fitness monitoring7.

Among various consumer devices incorporating optical HR monitoring, earbuds have emerged as an optimal location for integrating physiological signal sensors, offering many advantages over conventional wrist and finger recording sites8,9. For example, Since 2007, when research initially demonstrated the feasibility of collecting HR data from the ear10, subsequent studies have revealed that the quality of in-ear PPG signals is comparable to those acquired from the finger11,12. Furthermore, in-ear PPG offers a higher beat detection rate and more accurate HR estimation compared to rings13. From a physiological perspective, the ear canal can resist the impact of sweat on PPG signal quality during exercise. Additionally, the ear probe is situated near the carotid artery that supplies blood to the brain, it can reduce the delay in measuring blood oxygen levels14. However, it is still unknown whether earbuds can accurately measure HR changes during stationary cycling without being affected by motion artifacts and low perfusion issues.

Apart from this, despite the significant expansion in the market share of commercial HR earbuds in recent years, the bulk of research has predominantly focused on validating the efficacy of non-commercial in-ear devices. For instance, studies have scrutinized disparities in validity between wakefulness and sleep states13, as well as the precision during surgical procedures15. While a few investigations have delved into the performance of specific commercial earbuds during stationary cycling exercises, notable variances persist across different brands and devices9. Furthermore, studies reveal that the accuracy of HR measurements using wrist-worn PPG devices deteriorates as exercise intensity escalates16. Moreover, various types of activities can impact the accuracy of HR measurements17,18. Thus, it is imperative to assess the effectiveness of HR measurements across diverse activities and intensities.

In summary, the rigorous scientific validation of commercial devices, particularly for research purposes, is paramount before their widespread adoption. Despite previous evaluations of in-ear devices for heart rate (HR) measurement, Hera Leto—a state-of-the-art in-ear device utilizing PPG technology—lacks specific validity assessments. Therefore, this study aims to assess the validity of Hera Leto in measuring HR across various activities and intensities by employing the Polar H10 as a standard reference. Additionally, the study seeks to compare the HR measurement accuracy of Hera Leto with that of the wrist-worn PPG device Polar Vantage V and the arm-worn PPG device Polar Verity Sense..

Methods

Participants

Twenty-eight national first-level athletes, including twelve females and sixteen males specializing in track and field and skiing disciplines, were successfully recruited for this study, as outlined in Table 1. The sample size was guided by similar studies, where a comparable number of participants was deemed adequate for evaluating the validity of HR monitoring technologies19,20. All participants were nonsmokers, had a normal BMI, and reported no use of medications that could influence heart rate or exercise performance, such as beta blockers or antiarrhythmic agents. Female participants confirmed they were not using hormonal contraceptives and reported a consistent menstrual cycle length of 28 ± 2 days. Prior to the study, participants were fully briefed on the procedures, along with the associated risks and benefits, and provided their informed consent. The study protocol was approved by the Conjoint Health Research Ethics Board at Beijing Sport University (NO. 2024016 H), adhering to the standards of the Declaration of Helsinki.

Table 1 Participants’ characteristics and ramp-incremental test responses.
Full size table

Study design

Participants in this study were required to visit 12 to 13 times to the laboratory over an 8-week period, ensuring a minimum of 48 h between each visit and tests scheduled consistently at specific times of the day. During the initial visit, participants first underwent assessments of height, weight, and body composition, followed by a ramp-incremental exercise test. Body composition was measured using bioelectrical impedance analysis (InBody 230, InBody, Seoul, Korea). Based on the results of the incremental test, appropriate intensities were determined for four steady-state exercises, six severe intensity exhaustive exercises, and three high-intensity interval training (HIIT) sessions. All experimental trials were conducted using a cycle ergometer (Monark 839E, Monark, Vansbro, Sweden) in a climate-controlled room, maintaining temperatures between 18 °C and 21 °C and humidity levels between 50% and 60%. Participants were instructed to avoid consuming any food or caffeine-containing beverages for at least 2 and 8 h, respectively, prior to each visit, and to refrain from intense physical activity for 24 h before each session. During all tests, metabolic gases were collected using a gas analyzer (MetaMax 3B, Cortex Biophysic, Leipzig, Germany). The analyzer was calibrated before each test according to the manufacturer’s guidelines, including calibration for pressure, gas concentrations (standard gases: O2 = 15.00%, CO2 = 5.00%), and volume (3 L syringe).

Devices

Polar H10

The Polar H10 served as the criterion device in this study, showcasing its validity with a strong correlation of \(\:r\)=0.997 in comparison to ECG21. As a result, the Polar H10 has become a standard reference in validating the accuracy of numerous commercial HR monitoring devices17. The HR sensor was attached to a Polar Pro HR strap, which was positioned over the sternum. Real-time HR data from the Polar H10 were transmitted to the Polar Flow app, which captured HR readings at 1-second intervals.

Polar verity sense

The Polar Verity Sense was secured to the nondominant forearm using an armband, in accordance with the manufacturer’s instructions. The sensor was placed on the upper forearm or upper arm, with the lens facing inward and positioned firmly against the skin on the underside of the armband. The strap notch used for securing the band was recorded and replicated across all visits to ensure consistent placement and tightness. Care was taken to ensure a snug yet comfortable fit, avoiding excessive pressure while minimizing sensor movement during exercise. The device was set to “Indoor Cycling” mode using the Polar Flow app, activating the heart rate sensor mode for continuous real-time Bluetooth streaming. HR data were captured at 1-second intervals through six light-emitting diode sensors and transmitted live via Bluetooth to a smartphone equipped with the Polar Flow app (Polar Electro Oy). Upon completion of each visit, the data were uploaded to the Polar Flow web service (Polar Electro Oy).

Polar vantage V

The Polar Vantage V was securely attached to the nondominant wrist, positioned one finger width above the ulnar styloid process, according to the manufacturer’s instructions. The wristband was fastened firmly to ensure that the back sensor remained in constant contact with the skin and that the device did not shift during movement. The strap notch position was recorded and consistently used across all visits. Placement was verified by trained staff using a fit-check protocol recommended by the manufacturer: gently pressing both sides of the band to confirm that no green LED light was visible and that the sensor stayed in full contact with the skin. The device was set to “Indoor Cycling” mode using the device’s onboard interface to standardize LED intensity and HR recording settings. HR data were recorded at a frequency of 1-second intervals. The device utilizes Polar Precision Prime sensor fusion technology for HR measurement. After each visit, the collected data were uploaded to the Polar Flow web service (Polar Electro Oy).

Hera Leto

Hera Leto is developed and commercially sold by Actywell Digital Limited based in Hong Kong, China. The earbuds utilize multi-LED and multi-photodiode technology and employ multi-wavelength PPG (green light: 940 nm; red light: 625 nm) to measure pulse rate within the ear canal through reflection. In contrast, the specific wavelength used by the green PPG sensors in the Polar Verity Sense and Polar Vantage V has not been publicly disclosed by the manufacturer. Hera Leto is an in-ear earbud designed to fit snugly in the right ear, with the Ear-Gel nozzle pointing downward as recommended in the manual. The earbuds come with three sizes of sports Ear-Gel (S, M, L) for personal choice. During data collection, the Hera Leto HR app is used to acquire data, which is stored in files named according to the format: < device id> __YYYY_MM_DD_hh_mm_ss_< file_type>.csv, with a time resolution of 1-second intervals.

Study procedures

Ramp-incremental exercise

The ramp incremental test aims to assess maximal oxygen uptake (VO2 max), identify the first ventilatory threshold (VT1), the second ventilatory threshold (VT2) and determine the maximum power output (Pmax) reached during the test22. Ventilatory thresholds were determined using the online platform https://www.exercisethresholds.com, which integrates multiple gas exchange-based detection methods. Two investigators independently identified the thresholds, and results were confirmed by consensus. During their initial visit to the laboratory, participants underwent a 4-minute warm-up cycling at 50 W, followed by an incremental ramp test with a 30 W per minute increase until exhaustion. At the start of the test, participants selected a cadence between 60 and 90 bpm and were instructed to maintain this cadence consistently throughout the test. This selected cadence was also required for all subsequent tests. The ramp test concluded when a participant voluntarily stopped due to exhaustion or when they were unable to sustain their chosen cadence for more than five seconds, despite receiving significant encouragement. This exhaustion criterion remained consistent across all tests designed to assess the limits of exhaustion.

Steady-state exercise

Participants in this study undertook four stages of constant load exercises at intensities corresponding to 45%, 55%, 65%, and 75% of the power output at VO2max (Pmax), each stage lasting 10 min. The participants were divided into two groups for laboratory testing, with one group randomly assigned to the 45% and 75% Pmax conditions, and the other to the 55% and 65% Pmax conditions. Following completion of each intensity level, participants engaged in active rest periods lasting approximately 20–30 min before proceeding to the subsequent level. The average heart rate (HR) during the 8 th to 10 th minute of each load level was computed and then linearly correlated with the respective load intensity (power).

Severe intensity exhaustive exercise

Each participant was required to complete at least three constant load tests to exhaustion. These tests began with an initial 4-minute warm-up at 25 watts, immediately followed by an increase to a predetermined power output (PO). Based on the Pmax determined during the maximal oxygen uptake test, intensities are set at VT2, 100%, 110%, 130%, 150%, and up to 170% of the Pmax. Participants were provided verbal encouragement to maintain maximum effort throughout the exercise.

High intensity intermittent training

We conducted three types of intermittent exercises, each lasting a total of 4 min with a fixed 10-second rest interval between activity bouts. The first type consisted of 10 s of activity followed by 10 s of rest, repeated 12 times (HIIT10/10). The second type involved 20 s of activity followed by 10 s of rest, repeated 8 times (HIIT20/10). The third type comprised 30 s of activity followed by 10 s of rest, repeated 6 times (HIIT30/10). The exercise intensities were set at 100% of VO2max for HIIT10/10, 90%VO2max for HIIT20/10, and 85%VO2max for HIIT30/10, respectively.

Heart rate kinetics

This study focused on examining the differences in HR kinetics during various phases of exercise, specifically the ON phase (during exercise) and the OFF phase (during recovery). The analysis was limited to steady-state exercise and post-exercise periods due to the unsuitability of current kinetic equations for extreme exercise conditions23. The HR kinetics during exercise followed a pattern similar to that of VO2, albeit with noticeable differences24. The HR kinetics during the exercise (ON phase) was modeled using two exponential terms: \(\:\text{H}\text{R}\:\left(\text{t}\right)=\text{H}\text{R}\text{b}\text{a}\text{s}\text{e}\text{l}\text{i}\text{n}\text{e}+\text{A}\times\:(1-{e}^{-\text{t}/{\uptau\:}1})\). Consequently, the HR during recovery (OFF phase) was described by a model incorporating two exponential terms as well: \(\:\text{H}\text{R}\:\left(\text{t}\right)=\text{H}\text{R}\text{E}\text{n}\text{d}-\text{E}\text{x}+\text{A}\times\:(1-{e}^{-(\text{t}-\text{T}\text{D})/{\uptau\:}1})\). HRbaseline and HREnd-Ex represent the starting heart rates at the beginning of exercise, while TD accounts for the time delay before recovery begins. A represents the amplitude of heart rate change, and τ1 is the time constant that determines the rate of recovery—the smaller the value, the faster the recovery.

Data analysis

Data processing and analysis in this study were conducted in accordance with the HR validity verification guidelines published by the Towards Intelligent Health and Well-Being: Network of Physical Activity Assessment (INTERLIVE)6. Validity comparisons were made between second-by-second data from Hera Leto, Polar Verity Sense, and Polar Vantage V against data from Polar H10.

Statistical analyses were performed using SPSS 21.0, with results presented as mean ± standard deviation (\(\:x\) ± \(\:s\)). The normality of the data was tested using the Shapiro-Wilk test. Differences in HR measurements for HRVT1, HRVT2, and the maximum HR (HRmax) across devices were assessed using one-way repeated measures ANOVA, with pairwise comparisons conducted using the Bonferroni post hoc test. A significance level of 0.05 was set for statistical tests. Prior to analysis, all data underwent missing value testing; datasets with less than 10% missing or anomalous data points were retained16.

The relevant libraries were imported in PyCharm to facilitate data processing, mathematical operations, and statistical analyses, including Bland-Altman and repeated measures correlation analyses. Using data from four devices, we fitted a linear equation between HR and power. The accuracy and quality of these fitted equations were assessed using the Standard Error of Estimate (SEE), where a lower SEE value indicates a better fit, reflecting a closer alignment between the observed HR and the model’s predictions. Furthermore, these libraries were used to compute evaluation metrics such as Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Concordance Correlation Coefficient (CCC) for assessing the effectiveness of Hera Leto, Polar Verity Sense, and Polar Vantage V6. Finally, Excel was employed for calculating dropout rate. For validity thresholds, we used a MAPE value of less than 5% and a CCC greater than 0.9025.

Results

Validity of HR measurement across all activities and intensities

Hera Leto demonstrated high validity in HR measurement across different exercise intensities, as shown in Table 2. While Verity Sense and Vantage V performed comparably under steady-state and HIIT conditions, their HR readings showed greater deviations during severe-intensity constant load exercise (≥ 130% Pmax), particularly during the ON phases.

Table 2 Mean and standard deviation in each intensity level.
Full size table

Table 3 presents the aggregated validity metrics across all exercise intensities. Hera Leto exhibited the lowest overall MAE and MAPE values and the highest CCC, compared to Polar Verity Sense and Polar Vantage V. However, as shown in Table 2, the differences in device performance varied across specific exercise conditions, with comparable accuracy observed among the devices during steady-state and HIIT exercises.

Table 3 Validity of HR measurement with Polar H10 and other devices.
Full size table

In the Bland-Altman plots, the Hera Leto had lowest bias (ON=−2.24 bpm, OFF=−0.15 bpm, ALL=−0.67 bpm) in estimating HR compared to the other device (Polar Verity Sense: ON=−5.54 bpm, OFF=−0.22 bpm, ALL=−1.56 bpm; Polar Vantage V: ON=−4.34 bpm, OFF = 0.52 bpm, ALL=−0.70). Hera Leto (ON: \(\:\mathbf{r}\) = 0.970, P < 0.001, OFF: \(\:\mathbf{r}\) = 0.957, P < 0.001, ALL: \(\:\mathbf{r}\) = 0.970, P < 0.001) had the strongest correlation with Polar H10 in estimating HR compared to other devices (Polar Verity Sense, ON: \(\:\mathbf{r}\) = 0.781, P < 0.001, OFF: = 0.887, P < 0.001, ALL: \(\:\mathbf{r}\) = 0.878, P < 0.001; Polar Vantage V:, ON: \(\:\mathbf{r}\) = 0.865, P < 0.001, OFF: \(\:\mathbf{r}\) = 0.933, P < 0.001, ALL: \(\:\mathbf{r}\) = 0.927, P < 0.001) (Fig. 1).

Fig. 1
figure 1

Using Bland-Altman analysis with limits of agreement and repeated-measures correlation to compare HR from the Polar H10 with other devices during ON and OFF phases. For visualization purposes, 1% of the total second-by-second HR data was randomly sampled for the Bland-Altman plots, while all paired observations were used for statistical analysis.

Full size image

Validity of HR measurement across various intensities

As exercise intensity increases, the VT2 serves as a distinct demarcation. Below this intensity, the accuracy of measurements from Hera Leto, Polar Verity Sense, and Polar Vantage V is very high, with very similar MAE, MAPE, and CCC. However, above this intensity, the effectiveness of HR measurements from all devices decreases, although Hera Leto shows the slowest rate of decline, but it has the highest dropout rate (Fig. 2).

Fig. 2
figure 2

The validity of an HR measuring device varies with intensity, with darker colors representing higher values and lighter colors indicating lower values for MAE, MAPE, CCC, and Dropout Rate.

Full size image

HR response for specific exercise

Ramp-incremental exercise

Table 4 shows that during incremental load exercise, Hera Leto, Polar Verity Sense, and Polar Vantage V exhibit consistent performance in terms of MAE, MAPE, and CCC; however, Hera Leto generally has a higher dropout rate.

Table 4 Validity of HR measurement in ramp-incremental exercise.
Full size table

The Polar Verity Sense, Polar Vantage V, and Hera Leto show significant differences in measurements of HRpeak and HR-VT2 when compared to the Polar H10 (P < 0.05). The differences in HRpeak are respectively 2 (95% CI: 1 to 3) bpm, 4 (95% CI: 2 to 6) bpm, and 2 (95% CI: 0 to 4) bpm. For HRVT2, the differences are also 2 (95% CI: 1 to 3) bpm, 4 (95% CI: 2 to 6) bpm, and 2 (95% CI: 0 to 4) bpm. However, there are no significant differences in the measurements of HR-VT1 between these devices and the H10. Additionally, there are no significant differences in the OFF-phase kinetic parameters measured during Ramp-Incremental Exercise between the Polar Verity Sense, Polar Vantage V, Hera Leto, and the Polar H10 (P > 0.05) (Table 5).

Table 5 HRpeak, HR-VT1 and HR-VT2 across the HR devices.
Full size table

Steady-state exercise

Figure 3 demonstrates a high level of consistency in the linear equations for steady-state HR and power fitted using four devices. Table 6 displays the results of HR kinetics during steady-state exercise; as intensity increases, there are no significant differences in HR kinetics parameters between Hera Leto and Polar H10 (P > 0.05). However, there are significant differences between the Polar H10 and both Polar Verity Sense and Polar Vantage V in the parameters\(\:\:A\) or \(\:\tau\:1\) (P < 0.05). Additionally, there are no significant differences in the OFF-phase kinetic parameters of steady-state exercise measured at different intensities between Polar Verity Sense, Polar Vantage V, Hera Leto, and Polar H10 (\(\:P\) > 0.05).

Fig. 3
figure 3

The linear equations of HR versus power measured by different devices.

Full size image
Table 6 Fitting parameters for HR kinetics during the ON phase of steady-state exercise.
Full size table

Severe intensity exhaustive exercise

Figure 4 illustrates the OFF-phase kinetics for severe intensity exhaustive exercise. As the intensity increases, there are no significant differences in the kinetic parameters between Hera Leto and Polar H10. However, Polar Verity Sense and Polar Vantage V struggle to accurately capture the full HR kinetics, particularly the rapid recovery phase of HR.

Fig. 4
figure 4

Fitting Parameters A, TD, and τ1 for HR kinetics during the OFF phase of severe intensity exhaustive exercise.

Full size image

High intensity intermittent exercise

Table 7 shows that the validity of HR measurements by Hera Leto, Polar Verity Sense, and Polar Vantage V—assessed using MAPE and CCC—remained within acceptable limits across HIIT intervals.

Table 7 Validity of HR measurement in HIIT.
Full size table

Figure 5 illustrates the MAE, MAPE, CCC, and Mean ± SD of HR measurements during the intervals and rest periods of HIIT, taken by different devices. As the number of HIIT sets increases, both the HRs measured by the devices and their validity improve, with Hera Leto showing validity closest to that of the Polar H10. Additionally, there are no significant differences in the OFF-phase kinetic parameters of HIIT measured at different intensities between Polar Verity Sense, Polar Vantage V, Hera Leto, and Polar H10 (P > 0.05).

Fig. 5
figure 5

The validity of different devices in measuring HR during the intervals and rest periods of HIIT.

Full size image

Discussion

This study is the first to evaluate the accuracy of a commercial earbud device in monitoring heart rate (HR) during stationary cycling across different intensities and exercise types. Hera Leto showed relatively stable agreement with the criterion device (Polar H10) under varying conditions. In the recovery (OFF) phases of steady-state and HIIT sessions, its validity metrics (e.g., MAE, MAPE) remained within acceptable limits, comparable to those of Polar Verity Sense and Polar Vantage V. During the exercise (ON) phases, performance varied among devices, with larger errors emerging at higher intensities. In incremental load tests, all three devices remained within acceptable ranges overall, though accuracy declined at VO₂max and VT2. In constant load exercises, increasing intensity led to greater HR measurement errors across all devices, but Hera Leto maintained acceptable accuracy over a wider range. During low-volume HIIT, the validity of all three devices improved across intervals and generally remained acceptable. Notably, all devices exhibited consistent HR kinetics with Polar H10 during steady-state ON phases. However, in recovery phases, only Hera Leto showed HR kinetics closely matching those of Polar H10, while the other two devices underperformed, particularly during severe intensity exhaustive exercise.

Equipping with wearable devices to measure and further analyze physiological signals are gaining popularities in exercise26,27,28. Our research demonstrated that the validity of the Hera Leto, Polar Verity Sense, and Polar Vantage V devices deteriorates as exercise intensity increases. This pattern was confirmed across both incremental load intensity and different constant load intensities. Similar findings have been reported in several studies examining the validity of commercial wearable devices, noting an increase in HR measurement errors with rising exercise intensities20,29,30,31. For example, we observed that the HR measurements at VT1 from all three devices were consistent with the Polar H10, while discrepancies were noted at VT2 and VO2max. Interestingly, studies have indicated that the Apple Watch possesses good to very good criterion validity for measuring HRmax. We hypothesize that the differences may arise from the modalities used in the incremental load tests: the Apple Watch study utilized a treadmill, whereas our tests were conducted on a cycle ergometer. This is supported by findings from Muggeridge et al.16, who noted that compared to treadmill exercises, the validity of wrist-worn PPG devices was particularly poor during stationary cycling exercises. It’s worth noting that throughout the incremental load exercises, the HR validity measurements of Hera Leto, Polar Verity Sense, and Polar Vantage V all remained within acceptable ranges, consistent with the findings reported by Wallen et al.32. We only tentatively speculate that VT2 might represent the turning point where validity begins to decline, given the large intervals between our incremental intensities. Referring to Jo et al.29, who found this turning point to be at 116 bpm (ECG HR), our study suggests it may occur after the HR corresponding to VT1 (Polar H10: 139 ± 15). Lastly, notably, Hera Leto exhibits a later turning point for error reduction compared to Polar Verity Sense and Polar Vantage V. This suggests that ear-based devices like Hera Leto may offer advantages in sustaining HR measurement accuracy under increasing exercise intensities, particularly in high-performance and endurance training settings.

This study identified that Hera Leto demonstrated overall better validity during HIIT compared to Polar Verity Sense and Polar Vantage V. However, all three devices exhibited unacceptable measurement errors in the initial sets of HIIT. Across the three HIIT formats, the validity of HR measurements—assessed via MAPE, MAE, and CCC—remained within acceptable limits. However, a slight decline in accuracy was observed as interval duration increased (e.g., HIIT30/10), particularly in the CCC values of Verity Sense and Vantage V. We speculate that this phenomenon may be attributed to similar factors that lead to increased HR errors during extreme exercise. Firstly, during HIIT, participants tend to grip the bike handles tightly and contract upper arm muscles, which can introduce noise into HR estimation and reduce overall HR measurement validity19,31. Secondly, at the onset of HIIT, devices located on the arms, wrists, and in the ears may not be as sensitive to sudden changes in exercise intensity, resulting in lower validity of HR measurement. However, ear-based devices may exhibit higher sensitivity to changes in exercise intensity compared to the other two locations9. Lastly, the improved validity of all three devices with increasing sets may be attributed to rapid blood flow redistribution during rest periods. Conversely, longer interval HIIT, due to prolonged ischemic periods, may not achieve sufficient recovery within the same rest time, leading to persistent HR measurement errors. This also partly explains why all three devices exhibited poorer validity in measuring HR during sustained HIIT and sprint interval exercises16.

The evaluation of Hera Leto’s HR measurement using dynamic HR equations showed that it yielded more favorable validity metrics—such as lower MAPE and higher CCC—compared to Polar Verity Sense and Polar Vantage V, particularly during certain HIIT sessions. Additionally, we found that the HR kinetics during steady-state exercise and post-exercise, as measured by Hera Leto, Polar Verity Sense, and Polar Vantage V, could be fitted to HR dynamic equations. For the majority of activities, especially during rest, low to moderate intensity exercises, and post-exercise recovery phases, the HR measurement devices were generally effective18. Ear-based PPG devices, such as Hera Leto, hold potential as a solution to improve HR measurement validity during HIIT. This is because ear-based PPG devices typically employ multi-channel monitoring, which enhances the signal-to-noise ratio and coverage of PPG9. Moreover, during severe stress conditions, the body maximizes blood flow to maintain adequate cerebral perfusion, and in-ear PPG devices, being situated near the highly sensitive brain, can measure HR with minimal delay12. Although PPG is susceptible to constriction of peripheral blood flow in low temperatures, in-ear PPG has demonstrated resistance to these effects due to the unique characteristics of the ear canal33.

Limitations of the study

The validity of the Hera Leto device was comprehensively evaluated across various activity types and intensities, comparing its performance with commonly used PPG devices. However, this study has several limitations. One key limitation is the relatively small sample size, primarily due to the high dropout rate associated with the Hera Leto device, which reduced the number of paired samples available for effectiveness analysis. Additionally, the study lacks reliability data, making it difficult to draw definitive conclusions about the validity of the tested devices. While the controlled laboratory setting allowed for precise comparison across devices, it may not fully capture the complexity of real-world outdoor exercise scenarios, such as road cycling. Additionally, the study sample consisted of trained university athletes, which may limit generalizability to clinical populations or older individuals. Factors such as skin temperature and skin tone, which may influence PPG signal quality, were not specifically controlled and should be considered in future studies9. Moreover, the absence of detailed information on the engineering design and algorithmic processing of the Hera Leto device prevents a thorough analysis of the observed differences from a technical perspective. To minimize variability, all devices were kept on the same firmware version throughout the study, and automatic updates were disabled.