Performance of Wearable Pulse Oximetry During Controlled Hypoxia Induction
==========================================================================

* Yihang Jiang
* Connor Spies
* Ali R. Roghanizad
* Satasuk Joy Bhosai
* Laurie Snyder
* Ashley Burke
* David MacLeod
* Jessilyn Dunn

## Abstract

Oxygen saturation is a crucial metric used for monitoring patients with lung disease or other conditions who are at risk of hypoxemia. Recently, consumer wearable devices began incorporating oxygen saturation measurement tools, widening their use among individuals with lung disease. However, their performance has not been well characterized, particularly during blood oxygen desaturation, due to limitations in collecting comparison arterial blood gas measurements during exercise-induced hypoxemia. To address this knowledge gap, we performed a controlled oxygen desaturation study in nine healthy participants to mimic exercise-induced hypoxemia, reducing the blood oxygen saturation levels from 100% to 60% using a gas delivery system. We conducted a comparative analysis of the displayed oxygen saturation readings of both the Masimo MightySat Rx finger pulse oximeter and Apple Watch Series 7 with arterial blood oxygen saturation readings obtained from arterial blood gas samples. Both the Masimo MightySat Rx pulse oximeter and the Apple Watch Series 7 tended to overestimate the oxygen saturation. The Masimo pulse oximeter readings were more likely to fall within the acceptable error range than the Apple Watch (49.03% compared to 32.14%). Notably, both devices have limitations under low oxygen saturation levels (<88%) with an error of 1.96% and 4.99% for the Masimo and Apple Watch, respectively. Among the oxygen desaturation measurements at a desaturation rate larger than two percent per minute, a rate which is clinically correlated to sleep apnea, both devices increased their error slightly by 0.29% and 1.80% respectively for the Masimo and Apple Watch. However, no statistically significant increase was found.

**Author Summary** Blood oxygen saturation (SpO2) is an important measurement for monitoring patients with acute and chronic conditions that are associated with hypoxemia, including chronic obstructive pulmonary disease (COPD), heart failure, and sleep apnea. While smartwatches may provide a novel method for continuous and unobtrusive SpO2 monitoring, it is necessary to understand their accuracy and limitations to ensure that they are used in a fit-for-purpose manner, especially under conditions of low oxygen saturation.

## Introduction

Oxygen saturation represents how much hemoglobin is bound to oxygen. Measurement of oxygen saturation is considered the fifth vital sign to gain insight into a person’s health status alongside body temperature, blood pressure, pulse rate and respiratory rate(1). Hypoxemia, often defined as low oxygen saturation below 90%, is associated with symptoms such as shortness of breath, rapid breathing, and fast heartbeat. Early identification of an abnormal oxygen saturation values (2) is important in chronic obstructive pulmonary disease (COPD) patients where lung disease can lead to hypoxemia symptoms and lower oxygen saturation during exercise, resting or sleep (3–5). In COPD, oxygen desaturation during exercise can be caused by multiple factors such as ventilation-perfusion mismatching, diffusion-type limitation, and shunting(6). In a study with 60 normoxic moderate-to-stable COPD patients, 33 patients (55%) desaturated during a 6-minute-walk-test (6MWT), which is a standardized self-paced walking test that assesses the cardiovascular response during daily activities(7). For COVID-19 patients with similar exercise-induced desaturation who underwent a lung ultrasound scan, a significantly higher lung ultrasound score was found as compared with patients who didn’t have oxygen desaturation larger than 4% indicating worsening odds of poor outcomes as a result of COVID-19 infection(8,9).

Measurement of oxygen saturation can be classified into two methods: invasive and non-invasive(10). Arterial blood gas (ABG) analysis is an invasive oxygen saturation measurement method that requires the collection of arterial blood samples from an artery and measurement of arterial oxygen saturation (SaO2) by a dedicated co-oximeter machine. ABG analysis is very accurate and considered the ‘gold standard’, but ABG sample collection is painful and impractical for continuous monitoring(11,12). Pulse oximetry is based on the principle that the optical absorption of red and near-infrared light by oxyhemoglobin and deoxyhemoglobin differ and allow the non-invasive estimate of oxygen saturation (SpO2)(13). SpO2 is considered a reliable surrogate for SaO2, meeting the increasing need for home monitoring. Wearables devices have integrated technology to measure and display oxygen saturation, SpO2, which helped the global market value of smartwatches reach $33.5 billion in 2022(14). However, there is concern that there may be a limited accuracy of smartwatches to detect and identify low oxygen saturation, and that such failures may a result of technical errors(15) or human factors(16). A lack of reliability for low oxygen saturation detection would increase the risks for patients with hypoxemia (16–18).

Multiple research studies have used SaO2 as the gold standard measure of oxygen saturation and have focused on the bias of SaO2 from pulse oximeters (SpO2) or smartwatches without investigating the accuracy of pulse oximeters during a continuous oxygen desaturation process(19,20). Other research studies that used SpO2 or SaO2 as the reference standard during a controlled desaturation process did not discuss the relationship between the accuracy and desaturation rate(21,22). Our work(23) addresses this gap by analytically validating that pulse oximeters can accurately respond to a drop in SaO2 during a continuous oxygen desaturation protocol. We hypothesized that the accuracy of pulse oximeters and wearables would differ under hypoxemia and would also be affected by the oxygen desaturation rate. We employed a controlled gas delivery system to conduct the controlled desaturation protocol and change the SaO2 by changing the setting of end-tidal oxygen (PetO2) that participants breathe in (Fig 1). Oxygen saturation readings from the Apple Watch Series 7, acknowledged for its superior best performance among a selection of four commercial wearables(24,25), and Masimo MightySat Rx, an FDA-cleared pulse oximeter, were compared against the gold standard of ABG analysis.

![Fig 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/07/16/2024.07.16.24310506/F1.medium.gif)

[Fig 1.](http://medrxiv.org/content/early/2024/07/16/2024.07.16.24310506/F1)

Fig 1. 
Flow chart depicting the desaturation protocol tools for adjusting PetO2 and measuring SaO2. By adjusting the end-tidal PetO2 setting in the RespirAct RA-MR gas delivery system (A), the study participant will inhale gas with different oxygen, carbon dioxide and nitrogen concentrations through a fitted mask (B). Blood samples that were taken once the PetO2 was stable were run through the Radiometer ABL90 Flex blood gas analyzer (C) to measure the resulting SaO2 and PaO2.

## Results

### Study Population

Nine individuals, 5 males and 4 females, provided written informed consent to Duke IRB protocol Pro00105579 and completed the study. The median age was 25 years (range 19 - 28) and the median Body Mass Index (BMI) was 24.4 kg/m2 (range, 21.8-35.1 kg/m2). Participants ranged from Type II to Type VI on the Fitzpatrick skin tone scale (Table 1). On the ITA skin tone scale which defines ITA value larger than 50 degree as very light skin tone and ITA value smaller than -50 degree as very dark skin tone(26), participants ranged from -27° to 54°, with a median value of 7°.

View this table:
[Table 1.](http://medrxiv.org/content/early/2024/07/16/2024.07.16.24310506/T1)

Table 1. Patient demographics and characteristics

### Desaturation Protocol and ABG Analysis

The subjects underwent a controlled, stepwise descent of oxygen saturation from 100% to a target of 60%. This was achieved by a dedicated gas delivery system, RespirAct (Thornhill Research, Toronto, Canada) to target pre-specified end-tidal oxygen tensions, PetO2. In turn, the arterial partial pressure of oxygen saturation (PaO2) determines the arterial oxygen saturation (SaO2), which is measured by ABG and estimated non-invasively (SpO2) by the pulse oximeter and smartwatch devices. Under stable PetO2 conditions, we obtained serial ABG samples to measure SaO2.

Measurements were taken across the 9 study participants at 18 targeted PetO2 levels ranging from 32 to 250 mmHg, representing SaO2 from 60 to 100%. There were 308 SaO2 measurements from the ABG analysis.

### Device Performance

Three hundred and five out of 308 (99%) and 284 out of 308 (92%) possible SpO2 measurements were taken from Masimo MightySat Rx and Apple Watch Series 7, respectively. In other words, only 3 measurements were unable to be taken by the Masimo device, but 24 measurements could not be taken by the Apple Watch. The maximum SpO2 measurement from Masimo MightySat Rx was 100 percent, and all participants had SpO2 measurements that reached 100 percent. Likewise, the maximum SpO2 measurements from Apple Watch Series 7 was 100 percent and all participants had SpO2 measurements that reached 100 percent. The minimum SpO2 measurement from Masimo MightySat Rx was 53 percent, and minimum SpO2 measurements were below 65 percent for 7 out of 9 (78%) participants. However, the minimum SpO2 measurement from Apple Watch Series 7 was 61 percent, and the minimum SpO2 measurements were below 65 percent for only 2 out of 9 (22%) participants. The estimated SaO2 controlled by the gas delivery system is shown in Fig 2A and the SpO2 measurements of the two devices during the desaturation protocols are shown in Fig 2B and Fig 2C, where it can be seen that the Apple Watch measurements do not reach the same minimum values that the Masimo measurements do.

![Fig 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/07/16/2024.07.16.24310506/F2.medium.gif)

[Fig 2.](http://medrxiv.org/content/early/2024/07/16/2024.07.16.24310506/F2)

Fig 2. 
**(A)** Estimated SaO2 range using PetO2 setting and Hemoglobin Dissociation Curve. The green / red dots are the maximum / minimum estimated SaO2. There are three phases separated using solid black line: First oxygen desaturation sequence, oxygen saturation recovery and second oxygen desaturation sequence. **(B and C)** SpO2 measurements from the Masimo MightySat Rx (B) and the Apple Watch Series 7 (C) during the oxygen desaturation protocol (N=9 study participants).

### Missingness and accuracy of measurements

Each measurement from Masimo MightySat Rx and Apple Watch Series 7 was compared to the arterial blood gas (ABG) measurement to determine whether it fell within, above, or below the ABG measurement range. The percentage of measurements falling into these different categories demonstrates whether there exists a trend of over- or underestimation of SpO2 levels. Both the MightySat Rx and the Apple Watch Series 7 tended to overestimate SpO2 (Fig 3, Table 2). The Apple Watch Series 7 had a higher percentage of overestimated measurements (56.49% of SpO2 measurements were overestimated) as compared with the MightySat Rx (where 44.48% of measurements were overestimated). The Masimo MightySat Rx had a higher percentage of measurements (49.03%) falling within the accuracy range of the reference device compared to Apple Watch Series 7 (32.14%). Ultimately, both devices have a problem of overestimating SpO2, and the percentage of measurements that were overestimated was even higher than the percentage of measurements within the acceptable error range. In terms of missingness, the Apple Watch Series 7 had more missingness (7.79%) compared to the MightySat Rx (0.97%), indicating that both have high likelihood (>90%)for successfully obtaining a measurement when a measurement was attempted. The Apple Watch Series 7’s number of missing measurements (24 missed out of 308) was higher than the Masimo MightySat Rx which only had 3 missing measurements among 308 attempts.

![Fig 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/07/16/2024.07.16.24310506/F3.medium.gif)

[Fig 3.](http://medrxiv.org/content/early/2024/07/16/2024.07.16.24310506/F3)

Fig 3. 
Relative percentages of data falling into the categories of overestimated measurements (red), measurements within acceptable error ranges (blue), underestimated measurements (green), and missing measurements (grey) for the Masimo MightySat Rx and Apple Watch Series 7.

View this table:
[Table 2.](http://medrxiv.org/content/early/2024/07/16/2024.07.16.24310506/T2)

Table 2. Percentage and number of measurements falling in four categories

A Bland–Altman analysis demonstrated the difference between the SpO2 measurements from the Apple Watch and Masimo MightySat Rx and reference SaO2 measurements (Fig 4A, Fig 4B). The MDE of SpO2 for the Masimo MightySat Rx and the Apple Watch was 1.80% and 3.26%, respectively. This analysis demonstrates again the trend of overestimation for both devices and that the severity of overestimation by the Apple Watch Series 7 was higher than the overestimation by the Masimo MightySat Rx.

![Fig 4.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/07/16/2024.07.16.24310506/F4.medium.gif)

[Fig 4.](http://medrxiv.org/content/early/2024/07/16/2024.07.16.24310506/F4)

Fig 4. 
Bland-Altman plots demonstrate the differences between all simultaneous (A) SpO2 measurements of the smartwatch (Apple Watch Series 7), (B) oximeter (Masimo MightySat Rx) and SaO2 measurements from ABG analysis. Under hypoxemia (SaO2 < 88 %), it also demonstrates simultaneous SpO2 measurements of the (C) smartwatch (Apple Watch Series 7), (D) oximeter (Masimo MightySat Rx) and SaO2 measurements from ABG analysis. The solid line is the mean difference of the measurements (MDE). Dashed lines are the 95% limits of agreement.

### Accuracy of measurements under hypoxemia

Among the 308 measurements in the ABG analysis, 165 SaO2 measurements (54%) are below 88 percent, which is the clinical threshold for hypoxemia(27). Among these 165 valid SaO2 measurements, 164 (99%) and 140 (85%) SpO2 measurements were collected simultaneously from the Masimo MightySat Rx and Apple Watch Series 7 respectively.

A Bland–Altman analysis demonstrated the difference between the SpO2 measurements from each the Apple Watch and the Masimo MightySat Rx against ABG SaO2 measurements under hypoxemia. (Fig 4C, Fig 4D) The MDE of SpO2 for the Masimo MightySat Rx and the Apple Watch were 1.96% and 4.99%, respectively. This analysis demonstrates that the severity of the overestimation increases under hypoxemia. Also, it demonstrates that the device has lower accuracy during hypoxemia compared to normal blood oxygen conditions.

### Accuracy of measurements during desaturation

Among the 308 measurements in the ABG analysis, 102 SaO2 measurements are collected during the first oxygen desaturation sequence (Fig 2A, Phase 1) which consisted of 8 stable desaturation steps that were evenly spaced in time (see methods). Among these 102 valid SaO2 measurements, 101 and 92 SpO2 measurements were collected simultaneously from Masimo MightySat Rx and Apple Watch Series 7 respectively. Among the 101 and 92 SpO2 measurements, 42 out of 101 and 38 out of 92 measurements are taken when the oxygen desaturation rate (ODR) is larger than 2 percent per minute.

Welch’s t-test models was used to explore whether a statistically significant difference exists between measurements with an oxygen desaturation rate above or below two percent per minute– equivalent to the 4% oxygen desaturation index (ODI-4%) which is defined as the number of events with a more than 4% decrease in SpO2 within 120 seconds– has been found related to sleep apnea in multiple studies (28,29). When the ODR is larger than 2 percent per minute, the MDE forthe Masimo MightySat Rx increases from 1.60% to 1.89% and for the Apple Watch increases from 2.14% to 3.94%. (Fig 5) No significant correlation between ODR and the error of the Masimo MightySat Rx and Apple Watch Series 7 were found (p-values = 0.64 and 0.17, respectively), indicating that, while there is a trend of error increasing with ODR, this trend is not statistically significant.

![Fig 5.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/07/16/2024.07.16.24310506/F5.medium.gif)

[Fig 5.](http://medrxiv.org/content/early/2024/07/16/2024.07.16.24310506/F5)

Fig 5. 
Boxplots demonstrate the different errors of pulse oximetry between measurements with oxygen desaturation rate < 2%/min and >= 2%/min (A) SpO2 measurements of the Masimo MightySat Rx (B) SpO2 measurements of the smartwatch (Apple Watch Series 7)

## Materials and Methods

### Study Population

This study was conducted under the approval of the Duke University Health System (DUHS) Institutional Review Board and all participants signed informed consent before participating (Pro00105579 and Pro00110458). The criteria for study inclusion were healthy volunteers at least 18 years old. Exclusion criteria included individuals with peripheral vascular disease, Raynaud’s syndrome, cryoglobulinemia or any collagen vascular disease affecting the fingers, a history of blood clots in the last 6 months, an essential tremor, or gel nail polish or any other non-natural, non-removable discoloration of the forefinger.

### Devices and Data Collection

This study involved two sequential, stepwise oxygen desaturation sequences. In the first sequence the oxygen saturation (SaO2) was reduced in stepwise fashion from ∼100% to ∼60%, and then back to ∼100% over the course of approximately 60 minutes. The second sequence involved decreasing participants’ SaO2 from ∼100% to ∼80%, and then back to ∼100% over the course of 48 minutes. During the oxygen desaturation sequences, the displayed pulse oximeter saturation (SpO2) of two devices (the Apple Watch Series 7 and the Masimo MightySat Rx) were recorded at the time of arterial blood draws. Prior to beginning the desaturation protocol, skin tone of participants was measured using an objective colorimeter device (the Delfin SkinColorCatch) as well as visually by the study team using the Fitzpatrick scale (FP). The RespirAct RA-MR and the Radiometer ABL90 Flex were used respectively to alter and measure the SaO2. The RespirAct RA-MR is a sequential gas delivery system used to control oxygen saturation and provide steady-state plateaus prior to the blood gas sampling(30). The blood gas samples were then analyzed by the Radiometer ABL90 Flex co-oximeter to yield the arterial oxygen saturation (SaO2), considered to be the ‘gold standard’ measurement. Once a steady state of PetO2 and PetCO2 was achieved by the gas delivery system, arterial samples were drawn and analyzed using the Radiometer ABL90 Flex.

### Desaturation Protocol

Two measurements of SpO2 from each of the Apple Watch and Masimo MightySat were obtained at each of the oxygenation plateaus (where PetO2 was held at a steady state by the RespirAct RA-MR) to coincide with the timing of the arterial blood gas samples. During this procedure, participants were lying in the supine position on a standard hospital stretcher with an arterial catheter placed in their forearms. The first oxygen desaturation sequence (Figure 2A, Phase 1) involved 8 steps of approximately 6 minutes each with PetO2 set to 90, 60, 50, 60, 50, 37, 34, and 32 mmHg and estimated maximum SaO2 from Hemoglobin Dissociation Curve to be 100, 94, 89, 84, 79, 74, 69, and 64 percent. Following this sequence, PetO2 will be set to 250 mmHg for approximately 30 minutes to allow participants to recover their maximum oxygen saturation. The second oxygen desaturation sequence consisted of 8 steps of approximately 6 minutes each with PetO2 set to 90, 60, 50, 60, 50, 45, 40 and 250 mmHg, and estimated maximum SaO2 to be 100, 94, 89, 94, 89, 84, 79 and 100. The smartwatch was placed tightly on the right wrist, approximately one centimeter above the ulnar styloid process. Measurements on the watch were manually triggered when PetO2 reached steady state, and if the watch failed to produce a measurement, it was repositioned on the participant’s wrist and another measurement was attempted. If the device still failed to produce a measurement, no value was recorded for this specific time point and it was labeled as a missing (unobtainable) observation. The Masimo MightySat Rx was placed on the middle finger of the participant’s right hand, and recorded continuously at the rate of one reading per second during the study procedure.

### Metrics and statistical analysis

To assess the effectiveness of the pulse oximetry devices, we compared each SpO2 measurement from the Apple Watch and the Masimo MightySat Rx to the SaO2 measurements, determining if they fell within, above, or below the reference range (2%) which is considered the usual error range of most finger tip oximeters(1), or if they were missing. The Bland-Altman method was employed to assess the accuracy of SpO2 measurements across all readings, with separate analysis for readings where SaO2 value was below 88%. The threshold of 88% was chosen in line with the British Thoracic Society guidelines(27) indicating the point for intensive therapy to elevate oxygen saturation. Mean directional error (MDE) (Eq 1) and missingness (Eq 2) were utilized as metrics to gauge device performance. To evaluate device performance during oxygen desaturation, we focused on observations in the initial oxygen desaturation phase, where SaO2 settings transitioned uniformly from 100 percent to 60 percent. We utilized a linear mixed effects model to examine the impact of fixed effects from oxygen desaturation rate and random effects from individual participants on the difference between SpO2 and SaO2 measurements. ![Formula][1]</img>  

![Formula][2]</img> 

## Discussion

Wearables may be used for continuous measurements of oxygen saturation (SpO2) for COPD patients at home. To evaluate the performance of these devices under hypoxemia, we applied two oxygen desaturation sequences on nine subjects with controlled, stepwise descent of oxygen saturation from 100% to a target of 60%. We used the clinically important threshold of 88%, a decision point for administering supplemental oxygen, to evaluate MDE using Bland-Altman analysis. We compared the readings from the Masimo pulse oximeter and the Apple Watch Series 7 with ‘gold standard’ ABG readings under both low and normal oxygen saturation levels.

How best to evaluate the performance of pulse oximeters under hypoxemia remains an unsolved problem. In the US FDA’s guidance on pulse oximetry FDA assessment(31), it is stated that FDA-cleared pulse oximeters will have the lowest accuracy below 80%, but there is no standard to evaluate the performance during hypoxemia with rapid oxygen saturation value changes. Calculating the average bias between the ‘gold standard’ SaO2 and device-measured SpO2 over all values of blood oxygen saturation is insufficient to understand a tool’s performance as that performance may change at lower values or during rapid change(17).

Thus, evaluating pulse oximeters during hypoxemia requires a particularly challenging form of clinical validation (e.g., as defined in the V3 framework), involving observing decreases in oxygen saturation over both longer timescales (e.g., on the order of hours or days, which is more common in this field), as well as during rapid change (e.g. on the order of minutes). An example of a longer timescale clinical validation study occurred in COPD patients for one week; however, SpO2 was used as the reference standard instead of SaO2 because the study was performed in an ambulatory setting, which introduced potential in accuracies in the reference standard(32). We recommend that, in the evaluation of new tools for measuring blood oxygen saturation, it is important to 1) carefully evaluate the performance of the device under hypoxemic conditions at a variety of levels of low blood oxygenation, 2) specifically explore performance during hypoxemia through oxygen desaturation sequences over shorter timescales to determine how well the device can capture rapid changes in blood oxygen saturation, 3) use arterial blood gas measurements as the gold standard measurement against which to determine performance of the device, and 4) explore the accuracy of pulse oximeters during hypoxemia on people with different skin tones since it is reported black patients had nearly three times the frequency of occult hypoxemia (low oxygen levels in the blood) as detected by blood gas measurements but not detected by pulse oximetry(33).

### Limitations

The main limitation of this study is the small sample size, leading to insufficient representation across skin tone categories. Three out of the five Fitzpatrick skin tone categories only have one participant each, making it impossible to generalize our results to people with similar skin tones. Our study used wide coverage of oxygen saturation values since the target SpO2 ranged from 60% to 100%. However, this protocol cannot demonstrate SpO2 measurement performance over long periods since each participant only spent 60 minutes in the first oxygen desaturation sequence and 48 minutes in the second one. It is necessary to consistently monitor the oxygen saturation value overnight because nocturnal hypoxemia commonly occurs among COPD patients(34,35). It would be infeasible to induce oxygen desaturation for such a long period in healthy adults, which is why it is necessary to perform clinical validation among COPD patients with nocturnal desaturation, defined as having a SpO2 value below 90% for more than 30% of the time in bed during one or more nights(36,37). Future studies should recruit more subjects with at least 15% of darkly pigmented subjects guided by FDA(31) and prolong the monitoring process for the nocturnal desaturation caused by COPD or sleep apnea to verify the accuracy and variability of wearables.

## Data Availability

Deidentified data used for submission is available after acceptance.

## Acknowledgements

The content is solely the responsibility of the authors and does not necessarily represent the official views of AstraZeneca or the National Institutes of Health.

*   Received July 16, 2024.
*   Revision received July 16, 2024.
*   Accepted July 16, 2024.


*   © 2024, Posted by Cold Spring Harbor Laboratory

This pre-print is available under a Creative Commons License (Attribution 4.0 International), CC BY 4.0, as described at [http://creativecommons.org/licenses/by/4.0/](http://creativecommons.org/licenses/by/4.0/)

## References

1.  1.Pulse Oximetry > Fact Sheets > Yale Medicine [Internet]. [cited 2023 Dec 12]. Available from: [https://www.yalemedicine.org/conditions/pulse-oximetry](https://www.yalemedicine.org/conditions/pulse-oximetry)
    
    

2.  2.Vold ML, Aasebø U, Wilsgaard T, Melbye H. Low oxygen saturation and mortality in an adult cohort: the Tromsø study. BMC Pulm Med. 2015 Feb 12;15:9.
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

3.  3.Davis J. WebMD. [cited 2023 Dec 12]. What Is COPD Hypoxia? Available from: [https://www.webmd.com/lung/copd/copd-hypoxia](https://www.webmd.com/lung/copd/copd-hypoxia)
    
    

4.  4.Knower MT, Dunagan DP, Adair NE,  Chin J Robert. Baseline Oxygen Saturation Predicts Exercise Desaturation Below Prescription Threshold in Patients With Chronic Obstructive Pulmonary Disease. Archives of Internal Medicine. 2001 Mar 12;161(5):732–6.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/archinte.161.5.732&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=11231707&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000167339200014&link_type=ISI) 

5.  5.Little SA, Elkholy MM, Chalmers GW, Farouk A, Patel KR, Thomson NC. Predictors of nocturnal oxygen desaturation in patients with COPD. Respiratory Medicine. 1999 Mar 1;93(3):202–7.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0954-6111(99)90009-4&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=10464879&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000078766800010&link_type=ISI) 

6.  6.Full article: Exertional Desaturation in Patients with Chronic Obstructive Pulmonary Disease [Internet]. [cited 2024 May 16]. Available from: [https://www.tandfonline.com/doi/full/10.3109/15412550903341497#2b85d6ca-6520-4a3d-8e4a-aa9f2ee3f33d-b6de7b7c-de82-45a5-9538-313dd15c6659](https://www.tandfonline.com/doi/full/10.3109/15412550903341497#2b85d6ca-6520-4a3d-8e4a-aa9f2ee3f33d-b6de7b7c-de82-45a5-9538-313dd15c6659)
    
    

7.  7.Dogra AC, Gupta U, Sarkar M, Padam A. Exercise-induced desaturation in patients with chronic obstructive pulmonary disease on six-minute walk test. Lung India. 2015;32(4):320–5.
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

8.  8.Carlucci A, Paneroni M, Carotenuto M, Bertella E, Cirio S, Gandolfo A, et al. Prevalence of exercise-induced oxygen desaturation after recovery from SARS-CoV-2 pneumonia and use of lung ultrasound to predict need for pulmonary rehabilitation. Pulmonol. 2023 Dec 1;29:S4–8.
    
    

9.  9.Perrone T, Soldati G, Padovini L, Fiengo A, Lettieri G, Sabatini U, et al. A New Lung Ultrasound Protocol Able to Predict Worsening in Patients Affected by Severe Acute Respiratory Syndrome Coronavirus 2 Pneumonia. Journal of Ultrasound in Medicine. 2021 Aug;40(8):1627–35.
    
    

10. 10.Collins JA, Rudenski A, Gibson J, Howard L, O’Driscoll R. Relating oxygen partial pressure, saturation and content: the haemoglobin–oxygen dissociation curve. Breathe (Sheff). 2015 Sep;11(3):194–201.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1183/20734735.001415&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26632351&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

11. 11.Diagnostic accuracy of venous blood gases compared to arterial blood gases | European Respiratory Society [Internet]. [cited 2023 Dec 12]. Available from: [https://erj.ersjournals.com/content/42/Suppl\_57/P3987](https://erj.ersjournals.com/content/42/Suppl_57/P3987)
    
    

12. 12.Plüddemann A, Thompson M, Heneghan C, Price C. Pulse oximetry in primary care: primary care diagnostic technology update. Br J Gen Pract. 2011 May;61(586):358–9.
    
    [FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiRlVMTCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiYmpncCI7czo1OiJyZXNpZCI7czoxMDoiNjEvNTg2LzM1OCI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA3LzE2LzIwMjQuMDcuMTYuMjQzMTA1MDYuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

13. 13.Chan ED, Chan MM, Chan MM. Pulse oximetry: Understanding its basic principles facilitates appreciation of its limitations. Respiratory Medicine. 2013 Jun;107(6):789–99.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.rmed.2013.02.004&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=23490227&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

14. 14.Global Market Insights Inc. [Internet]. [cited 2023 Dec 12]. Smartwatch Market Size & Share – Analysis Report, 2032. Available from: [https://www.gminsights.com/industry-analysis/smartwatch-market](https://www.gminsights.com/industry-analysis/smartwatch-market)
    
    

15. 15.An assessment of the accuracy of pulse oximeters - Milner - 2012 - Anaesthesia - Wiley Online Library [Internet]. [cited 2023 Dec 12]. Available from: [https://associationofanaesthetists-publications.onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2044.2011.07021.x](https://associationofanaesthetists-publications.onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2044.2011.07021.x)
    
    

16. 16.Wilson BJ, Cowan HJ, Lord JA, Zuege DJ, Zygun DA. The accuracy of pulse oximetry in emergency department patients with severe sepsis and septic shock: a retrospective cohort study. BMC Emerg Med. 2010 May 5;10:9.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/1471-227X-10-9&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20444248&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

17. 17.Accuracy of Pulse Oximeters in Detecting Hypoxemia in Patients with Chronic Thromboembolic Pulmonary Hypertension - PMC [Internet]. [cited 2023 Dec 12]. Available from: [https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4433184/](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4433184/)
    
    

18. 18.Louie A, Feiner JR, Bickler PE, Rhodes L, Bernstein M, Lucero J. Four Types of Pulse Oximeters Accurately Detect Hypoxia during Low Perfusion and Motion. Anesthesiology. 2018 Mar 1;128(3):520–30.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/ALN.0000000000002002&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

19. 19.Rafl J, Bachman TE, Rafl-Huttova V, Walzel S, Rozanek M. Commercial smartwatch with pulse oximeter detects short-time hypoxemia as well as standard medical-grade device: Validation study. Digit Health. 2022 Oct 11;8:20552076221132127.
    
    

20. 20.Santos M, Vollam S, Pimentel MA, Areia C, Young L, Roman C, et al. The Use of Wearable Pulse Oximeters in the Prompt Detection of Hypoxemia and During Movement: Diagnostic Accuracy Study. Journal of Medical Internet Research. 2022 Feb 15;24(2):e28890.
    
    

21. 21.The desaturation response time of finger pulse oximeters during mild hypothermia* - MacLeod - 2005 - Anaesthesia - Wiley Online Library [Internet]. [cited 2024 Jun 9]. Available from: [https://associationofanaesthetists-publications.onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2044.2004.04033.x](https://associationofanaesthetists-publications.onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2044.2004.04033.x)
    
    

22. 22.Pulse Oximeter Performance during Rapid Desaturation - PMC [Internet]. [cited 2024 Jun 9]. Available from: [https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9185462/](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9185462/)
    
    

23. 23.Goldsack JC, Coravos A, Bakker JP, Bent B, Dowling AV, Fitzer-Attas C, et al. Verification, analytical validation, and clinical validation (V3): the foundation of determining fit-for-purpose for Biometric Monitoring Technologies (BioMeTs). npj Digit Med. 2020 Dec;3(1):55.
    
    

24. 24.Jiang Y, Spies C, Magin J, Bhosai SJ, Snyder L, Dunn J. Investigating the accuracy of blood oxygen saturation measurements in common consumer smartwatches. PLOS Digit Health. 2023 Jul 12;2(7):e0000296.
    
    

25. 25.Windisch P, Schröder C, Förster R, Cihoric N, Zwahlen DR. Accuracy of the Apple Watch Oxygen Saturation Measurement in Adults: A Systematic Review. Cureus. 15(2):e35355.
    
    

26. 26.Agarwal M, Anand S, Patro M, Gothi D. Early versus non-early desaturation during 6MWT in COPD patients: A follow-up study. Lung India. 2023;40(3):235–41.
    
    

27. 27.BTS guideline for emergency oxygen use in adult patients | Thorax [Internet]. [cited 2023 Dec 12]. Available from: [https://thorax.bmj.com/content/63/Suppl\_6/vi1](https://thorax.bmj.com/content/63/Suppl_6/vi1)
    
    

28. 28.Jung H, Kim D, Lee W, Seo H, Seo J, Choi J, et al. Performance evaluation of a wrist-worn reflectance pulse oximeter during sleep. Sleep Health. 2022 Oct 1;8(5):420–8.
    
    

29. 29.Temirbekov D, Güneş S, Yazıcı ZM, Sayın İ. The Ignored Parameter in the Diagnosis of Obstructive Sleep Apnea Syndrome: The Oxygen Desaturation Index. Turk Arch Otorhinolaryngol. 2018 Mar;56(1):1–6.
    
    

30. 30.Ikeda K, MacLeod DB, Grocott HP, Moretti EW, Ames W, Vacchiano C. The Accuracy of a Near-Infrared Spectroscopy Cerebral Oximetry Device and Its Potential Value for Estimating Jugular Venous Oxygen Saturation. Anesth Analg. 2014 Dec;119(6):1381–92.
    
    

31. 31.Health C for D and R. Pulse Oximeter Accuracy and Limitations: FDA Safety Communication. FDA [Internet]. 2023 Nov 17 [cited 2023 Dec 12]; Available from: [https://www.fda.gov/medical-devices/safety-communications/pulse-oximeter-accuracy-and-limitations-fda-safety-communication](https://www.fda.gov/medical-devices/safety-communications/pulse-oximeter-accuracy-and-limitations-fda-safety-communication)
    
    

32. 32.Buekers J, Theunis J, De Boever P, Vaes AW, Koopman M, Janssen EV, et al. Wearable Finger Pulse Oximetry for Continuous Oxygen Saturation Measurements During Daily Home Routines of Patients With Chronic Obstructive Pulmonary Disease (COPD) Over One Week: Observational Study. JMIR Mhealth Uhealth. 2019 06;7(6):e12866.
    
    

33. 33.Sjoding MW, Dickson RP, Iwashyna TJ, Gay SE, Valley TS. Racial Bias in Pulse Oximetry Measurement. N Engl J Med. 2020 Dec 17;383(25):2477–8.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMc2029240&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=33326721&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

34. 34.Dewan NA, Nieto FJ, Somers VK. Intermittent Hypoxemia and OSA. Chest. 2015 Jan;147(1):266–74.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1378/chest.14-0500&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25560865&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

35. 35.Oxygen saturation during daily activities in chronic obstructive pulmonary disease | European Respiratory Society [Internet]. [cited 2023 Dec 20]. Available from: [https://erj.ersjournals.com/content/9/12/2584.long](https://erj.ersjournals.com/content/9/12/2584.long)
    
    

36. 36.Iliaz S, Cagatay T, Bingol Z, Okumus G, Iliaz R, Kuran G, et al. Does the 6-minute walk test predict nocturnal oxygen desaturation in patients with moderate to severe COPD? Chron Respir Dis. 2015 Feb;12(1):61–8.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/1479972314562406&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25480424&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F16%2F2024.07.16.24310506.atom) 

37. 37.Chaouat A, Weitzenblum E, Kessler R, Charpentier C, Ehrhart M, Levi-Valensi P, et al. Sleep-related O2 desaturation and daytime pulmonary haemodynamics in COPD patients with mild hypoxaemia. Eur Respir J. 1997 Aug;10(8):1730–5.
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiZXJqIjtzOjU6InJlc2lkIjtzOjk6IjEwLzgvMTczMCI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA3LzE2LzIwMjQuMDcuMTYuMjQzMTA1MDYuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9)

 [1]: /embed/graphic-8.gif
 [2]: /embed/graphic-9.gif