In this study, 2 identical Apple Watch Series 8 devices (45 mm) were used to monitor the HR of surgeons during robotic-assisted surgery. The surgeons performing the robotic-assisted surgery wore an Apple Watch on each wrist and upper arm and used the controls on the console to maneuver the robotic arm on the patient-side cart. When the Apple Watch was attached to the upper arm, the band was lengthened using rubber bands to adjust to a position causing the least discomfort to the surgeon (Figure 1).

In this adjusted position, the watch was ideally located directly over the superficial vein of the upper arm. The readings were then compared to those from a fingertip pulse oximeter to ensure general consistency (Figure 2).

This study included monitoring the HR of surgeons during the first hour of console control. HR data were collected using the Hachi app provided by APTECH, which enabled the collection and extraction of HR data at 1-minute intervals from multiple Apple Watches on a single iPad (Apple). This app also facilitated centralized data management. The timing for initiating HR measurement within this 1-minute interval was not determined by the examiner’s discretion but was dependent on the app’s functionality. In addition to HR data, demographic information, such as gender, age, weight, height, and body mass index (BMI), was collected for all surgeons. Other information recorded included console time, operative time, and the type of surgical procedure performed. Wearing an Apple Watch on the upper arm is not a method recommended by Apple, and there are no reports evaluating the concordance of HR measurements between 2 Apple Watches simultaneously worn by the same individual at different anatomical sites. As an additional step to evaluate the results obtained in the study, HR data were also collected from a single surgeon by wearing both Apple Watches on both wrists using the same method. The purpose of this supplementary data collection was to check if the observed differences in HR measurements between the upper arm and wrist positions were within an acceptable range.

Informed consent was obtained from all the participants. This study was conducted in accordance with the ethical standards of the Helsinki Declaration of 1975. Ethics approval was obtained from the Ethics Committee of the Gunma Prefectural Cancer Center (405-04030).

The mean difference (MD) and SD of difference (SDD) between the wrist-worn and the upper arm–worn Apple Watches were calculated, and Bland-Altman plots were constructed to exclude systematic errors. Bias (MD) and limits of agreement (LoA; MD ± 1.96 × SDD) were plotted on the Bland-Altman plots to assess clinical applicability. The accuracy of the HR measurement from the Apple Watch worn on the upper arm was assessed based on the mean absolute error (MAE) and mean absolute percentage error (MAPE) between the upper arm and the wrist. MAE reflects the average size of the differences between predicted and observed values and ranges from zero to infinity, where lower MAE values indicate better forecasting performance. MAPE is commonly used as a measure of the prediction accuracy of a forecasting method. It is an average of the absolute values of the errors divided by the observed values. MAPE ranges from 0% to 100%, where lower MAPE values indicate better predictive performance of the model. In general, a MAPE of less than 10% is considered highly predictive [9].

The intraclass correlation coefficient (ICC) was calculated to determine the correlation between the Apple Watch on the upper arm and the one on the wrist. ICC estimates and their 95% CIs were calculated using SPSS statistical package version 22 (SPSS) based on a single rater (k=1), consistency, and a 2-way mixed-effects model.

Based on the 95% CI of the ICC estimate, values <0.5, between 0.5-0.75, between 0.75-0.9, and >0.90 were considered to be indicative of poor, moderate, good, and excellent reliability, respectively [9]. For all statistical tests, the alpha level adopted for significance (2-tailed) was set at P<.05.

The Bland-Altman plots, MAEs, MAPEs, and ICCs were the statistical measures used to evaluate the agreement and reliability of measurements in this research on wrist-worn devices capable of monitoring HR. The study found a small bias between the upper arm–worn and wrist-worn devices, no systematic error, and a high predictive value for MAPE and moderate predictive value for ICC for each participant. It was concluded that there is good agreement and reliability of the measurements obtained by the Apple Watch when comparing the upper arm–worn device with the wrist-worn device.

Unexpectedly, the least agreement between the 2 devices was found in the supplementary data involving a surgeon wearing an Apple Watch on each wrist in the correct manner. It was assumed that having the watches worn correctly on both wrists would provide the most accurate and reliable measurements. However, experiments where the watch was worn on upper arm and wrist on the same side showed that hand movements were generally consistent, enabling more stable measurements. In contrast, when the watch was worn on both wrists, the left and right hand movements were completely different, which could have resulted in a significant discrepancy in the measured values. In robotic surgery, where the robot’s arms can bend beyond the natural range of human wrist motion, surgeons often bend their wrists to the limit. We hypothesize that this extreme movement may reduce venous blood flow, thereby increasing the likelihood of discrepancies in HR measurements. It is important to note that this is a speculation, and further studies are needed to confirm the cause of this unexpected result. Nonetheless, this finding highlights the importance of understanding the limitations and potential sources of error when using wearable devices for health monitoring purposes.

HR is associated with survival in both healthy individuals and patients with various underlying cardiovascular diseases [10-12]. For example, a resting HR above 75 beats per minute in healthy individuals is known to increase the risk of sudden death from myocardial infarction [13]. Additionally, experiencing stressful life events increases the risk of developing cardiovascular disease [14].

An increase in HR leads to a decrease in diastolic time and an increase in systolic time, resulting in decreased myocardial perfusion and increased left ventricular work. These changes can ultimately lead to left ventricular hypertrophy, myocardial damage, and congestive heart failure. Increased HR may also be associated with endothelial damage, oxidative stress, inflammation, and vascular stiffness, which can contribute to aging, the development of atherosclerosis, arterial hypertension, and a stiff aorta. An increase in pacing rate from 60 to 90 beats per minute in humans has been shown to reduce the distensibility of the carotid and radial arteries [15]. Moreover, acute stress can cause sympathetic nervous system activation and parasympathetic nervous system suppression, leading to greater myocardial contraction and an increased HR. This can also cause an increased inflammatory (IL-6) response due to altered autonomic nervous system activity, which is associated with an increased risk of cardiovascular disease [16,17]. Given these associations, it is considered important to monitor the HR of health care professionals. However, the specific nature of their work makes this very difficult, and very little research has been done in this area.

In our experiment, we took care to position the Apple Watch directly above the blood vessels on the body surface when it was worn on the upper arm. We could confirm that there was never an instance in which HR measurement failed during the experiment when the watch was worn on either the upper arm or the wrist. While Apple Watch HR measurements have generally been found to be accurate, there are several factors known to cause significant errors in the readings. One of the most common factors is when the Apple Watch is not worn snugly on the wrist. Accuracy can also be significantly reduced during high-impact activities, such as running or cycling [18]. Additionally, the darker the skin tone, the less accurate the readings are. It has also been suggested that accuracy may be reduced in obese people due to increased subcutaneous fat thickness [18,19]. Contrasting previous literature that points toward higher BMI as a source of measurement errors, our study challenges this notion. Specifically, Participant A (BMI 24.7) and Participant B (BMI 30.3) yielded reliable HR measurements. In contrast, Participant C, with a lower BMI of 18, produced measurements that were somewhat less reliable when compared to the other participants. For instance, Participant C’s scant subcutaneous fat could have hindered the Apple Watch’s skin adherence, compromising the accuracy of measurements. This leads us to consider that both extremes of body composition—be it obesity or leanness—could challenge the reliability of wrist-worn HR monitors like the Apple Watch. Finally, the HR per minute was recorded simultaneously on 2 Apple watches, but the measurement times were dependent on the Apple Watch and could not be matched exactly. This resulted in a potential 28-second measurement error. This discrepancy in measurement times may be one of the reasons why the measurements did not match exactly.

The study suggests that the Apple Watch, worn on the upper arm, could be used to measure the HR of health care professionals in confined surgical environments without the need for disinfection. This would make mental and physical stress monitoring convenient and reliable. This study is the first to use an Apple Watch worn on the upper arm to measure the HR of surgeons during surgery. The findings suggest that wearable devices, such as the Apple Watch, could be used to measure the HR of health care workers during surgical procedures where there are limitations in measuring vital signs. This can enable an analysis of specific time periods and provide a more focused understanding of how HR is affected during this critical period of the surgical procedure. However, it is important to note that the feasibility of using the upper arm placement may be compromised in activities requiring extensive movement. In such scenarios, the device may become dislodged, thereby affecting the reliability of HR measurements. Consequently, we recommend reserving upper arm mounting for specific, controlled environments, such as surgeries that involve a limited range of motions, similar to those associated with surgical operations.

This study has some limitations that should be considered. First, the sample size was small, and the inclusion of only 4 male Japanese doctors in robotic surgery may have led to selection bias. Therefore, the findings may not be generalizable to a wider population. Second, the method of wearing the Apple Watch on the upper arm for HR measurement is not recommended by Apple and was only evaluated within the limited range of movements during surgical procedures. Therefore, it may not be suitable for other types of physical activities or movements. Third, the timing of the measurements could not be exactly matched between the 2 Apple Watches, making the data less consistent.

Our study showed that the HR measurements obtained from an Apple Watch worn on the upper arm during robotic-assisted surgery were moderately correlated and consistent with the measurements obtained from an Apple Watch worn on the wrist. The MAE and MAPE between the 2 positions were low, indicating an acceptable level of correlation and a high level of accuracy. Our findings suggest that the upper arm is a viable alternative to the wrist for monitoring HR during surgery when it is not feasible to wear a watch on the wrist. These findings have important implications for improving data collection and management of the physical and mental demands of operating room staff during surgery, where wearing a watch on the wrist may not be feasible.