OBJECTIVE: This study aims to develop sleep apnea screening models with overnight SpO2 data, and to investigate the impact of the SpO2 data granularity on model performance.
METHODS: A total of 7,718 SpO2 recordings from the SHHS and MESA datasets were used. Probabilistic ensemble machine learning was employed to predict sleep apnea status at three AHI cutoff points: ≥ 5, ≥ 15, and ≥ 30 events/hour. To investigate the impact of data granularity, SpO2 data were aggregated at 30, 60, and 300 s.
RESULTS: Our models demonstrated good to excellent performance on internal test, with average area under the curve (AUC) values of 0.91, 0.93, and 0.96 for cutoffs ≥ 5, ≥ 15, and ≥ 30 at data granularity of 1 s, respectively. Both sensitivity (0.76, 0.84, 0.89) and specificity (0.87, 0.86, 0.90) ranged from good to excellent across three cutoffs. Positive predictive values (PPV) ranged from excellent to fair (0.97, 0.83, 0.66), and negative predictive values (NPV) ranged from low to excellent (0.43, 0.87, 0.98). Model performance on external test slightly dropped compared to internal test, but still achieved good to excellent AUC above 0.80 across all data granularity and all the three cutoffs. Data granularity of 300 s led to a reduction in performance metrics across all cutoffs.
CONCLUSIONS: Our models demonstrated superior performance across all three AHI cutoff thresholds compared to existing large sleep apnea screening models, even when considering varying SpO2 data granularity. However, lower data granularity was associated with decreased screening performance, indicating a need for further research in this area.

BACKGROUND: Many commodity pulse oximeters are insufficiently calibrated for patients with darker skin. We demonstrate a quantitative measurement of this disparity in peripheral blood oxygen saturation (SpO2) with a controlled experiment. To mitigate this, we present OptoBeat, an ultra-low-cost smartphone-based optical sensing system that captures SpO2 and heart rate while calibrating for differences in skin tone. Our sensing system can be constructed from commodity components and 3D-printed clips for approximately US $1. In our experiments, we demonstrate the efficacy of the OptoBeat system, which can measure SpO2 within 1% of the ground truth in levels as low as 75%.
OBJECTIVE: The objective of this work is to test the following hypotheses and implement an ultra-low-cost smartphone adapter to measure SpO2: skin tone has a significant effect on pulse oximeter measurements (hypothesis 1), images of skin tone can be used to calibrate pulse oximeter error (hypothesis 2), and SpO2 can be measured with a smartphone camera using the screen as a light source (hypothesis 3).
METHODS: Synthetic skin with the same optical properties as human skin was used in ex vivo experiments. A skin tone scale was placed in images for calibration and ground truth. To achieve a wide range of SpO2 for measurement, we reoxygenated sheep blood and pumped it through synthetic arteries. A custom optical system was connected from the smartphone screen (flashing red and blue) to the analyte and into the phone\'s camera for measurement.
RESULTS: The 3 skin tones were accurately classified according to the Fitzpatrick scale as types 2, 3, and 5. Classification was performed using the Euclidean distance between the measured red, green, and blue values. Traditional pulse oximeter measurements (n=2000) showed significant differences between skin tones in both alternating current and direct current measurements using ANOVA (direct current: F2,5997=3.1170 × 105, P<.01; alternating current: F2,5997=8.07 × 106, P<.01). Continuous SpO2 measurements (n=400; 10-second samples, 67 minutes total) from 95% to 75% were captured using OptoBeat in an ex vivo experiment. The accuracy was measured to be within 1% of the ground truth via quadratic support vector machine regression and 10-fold cross-validation (R2=0.97, root mean square error=0.7, mean square error=0.49, and mean absolute error=0.5). In the human-participant proof-of-concept experiment (N=3; samples=3 × N, duration=20-30 seconds per sample), SpO2 measurements were accurate to within 0.5% of the ground truth, and pulse rate measurements were accurate to within 1.7% of the ground truth.
CONCLUSIONS: In this work, we demonstrate that skin tone has a significant effect on SpO2 measurements and the design and evaluation of OptoBeat. The ultra-low-cost OptoBeat system enables smartphones to classify skin tone for calibration, reliably measure SpO2 as low as 75%, and normalize to avoid skin tone-based bias.

Wearable devices have been used extensively both inside and outside of the hospital setting. During the COVID-19 pandemic, in some contexts, there was an increased need to remotely monitor pulse and saturated oxygen for patients due to the lack of staff and bedside monitors.
A prototype of a remote monitoring system using wearable pulse oximeter devices was implemented at the Hospital for Tropical Diseases in Ho Chi Minh City, Vietnam, from August to December 2021. The aim of this work was to support the ongoing implementation of the remote monitoring system.
We used an action learning approach with rapid pragmatic methods, including informal discussions and observations as well as a feedback survey form designed based on the technology acceptance model to assess the use and acceptability of the system. Based on these results, we facilitated a meeting using user-centered design principles to explore user needs and ideas about its development in more detail.
In total, 21 users filled in the feedback form. The mean technology acceptance model scores ranged from 3.5 (for perceived ease of use) to 4.4 (for attitude) with behavioral intention (3.8) and perceived usefulness (4.2) scoring in between. Those working as nurses scored higher on perceived usefulness, attitude, and behavioral intention than did physicians. Based on informal discussions, we realized there was a mismatch between how we (ie, the research team) and the ward teams perceived the use and wider purpose of the technology.
Designing and implementing the devices to be more nurse-centric from their introduction could have helped to increase their efficiency and use during the complex pandemic period.

UNASSIGNED: Sleep apnea is a common sleep disorder affecting a significant portion of the population, but many apnea patients remain undiagnosed because existing clinical tests are invasive and expensive. This study aimed to develop a method for easy sleep apnea screening.
UNASSIGNED: Three supervised machine learning algorithms, including logistic regression, support vector machine, and light gradient boosting machine, were applied to develop apnea screening models at two apnea-hypopnea index cutoff thresholds: ≥ 5 and ≥ 30 events/hours. The SpO2 recordings of the Sleep Heart Health Study database (N = 5786) were used for model training, validation, and test. Multiscale entropy analysis was performed to derive a set of multiscale attention entropy features from the SpO2 recordings. Demographic features including age, sex, body mass index, and blood pressure were also used. The dependency among the multiscale attention entropy features were handled with the independent component analysis.
UNASSIGNED: For cutoff ≥ 5/hours, logistic regression model achieved the highest Matthew\'s correlation coefficient (0.402) and area under the curve (0.747), and reasonably good sensitivity (75.38%), specificity (74.02%), and positive predictive value (92.94%). For cutoff ≥ 30/hours, support vector machine model achieved the highest Matthew\'s correlation coefficient (0.545) and area under the curve (0.823), and good sensitivity (82.00%), specificity (82.69%), and negative predictive value (95.53%).
UNASSIGNED: Our models achieved better performance than existing methods and have the potential to be integrated with home-use pulse oximeters.

BACKGROUND: Carbon monoxide (CO) poisoning is an important cause of morbidity and mortality worldwide. Symptoms are mostly aspecific, making it hard to identify, and its diagnosis is usually made through blood gas analysis. However, the bulkiness of gas analyzers prevents them from being used at the scene of the incident, thereby leading to the unnecessary transport and admission of many patients. While multiple-wavelength pulse oximeters have been developed to discriminate carboxyhemoglobin (COHb) from oxyhemoglobin, their reliability is debatable, particularly in the hostile prehospital environment.
OBJECTIVE: The main objective of this pilot study was to assess whether the Avoximeter 4000, a transportable blood gas analyzer, could be considered for prehospital triage.
METHODS: This was a monocentric, prospective, pilot evaluation study. Blood samples were analyzed sequentially with 2 devices: the Avoximeter 4000 (experimental), which performs direct measurements on blood samples of about 50 µL by analyzing light absorption at 5 different wavelengths; and the ABL827 FLEX (control), which measures COHb levels through an optical system composed of a 128-wavelength spectrophotometer. The blood samples belonged to 2 different cohorts: the first (clinical cohort) was obtained in an emergency department and consisted of 68 samples drawn from patients admitted for reasons other than CO poisoning. These samples were used to determine whether the Avoximeter 4000 could properly exclude the diagnosis. The second (forensic) cohort was derived from the regional forensic center, which provided 12 samples from documented CO poisoning.
RESULTS: The mean COHb level in the clinical cohort was 1.7% (SD 1.8%; median 1.2%, IQR 0.7%-1.9%) with the ABL827 FLEX versus 3.5% (SD 2.3%; median 3.1%, IQR 2.2%-4.1%) with the Avoximeter 4000. Therefore, the Avoximeter 4000 overestimated COHb levels by a mean difference of 1.8% (95% CI 1.5%-2.1%). The consistency of COHb readings by the Avoximeter 4000 was excellent, with an intraclass correlation coefficient of 0.97 (95% CI 0.93-0.99) when the same blood sample was analyzed repeatedly. Using prespecified cutoffs (5% in nonsmokers and 10% in smokers), 3 patients (4%) had high COHb levels according to the Avoximeter 4000, while their values were within the normal range according to the ABL827 FLEX. Therefore, the specificity of the Avoximeter 4000 in this cohort was 95.6% (95% CI 87%-98.6%), and the overtriage rate would have been 4.4% (95% CI 1.4%-13%). Regarding the forensic samples, 10 of 12 (83%) samples were positive with both devices, while the 2 remaining samples were negative with both devices.
CONCLUSIONS: The limited difference in COHb level measurements between the Avoximeter 4000 and the control device, which erred on the side of safety, and the relatively low overtriage rate warrant further exploration of this device as a prehospital triage tool.

UNASSIGNED: This study aimed to investigate (1) the role of mouth puffing phenomenon and upper airway features in obstructive sleep apnea (OSA) and (2) whether mouth-taping during sleep alleviated the severity of OSA.
UNASSIGNED: Seventy-one participants underwent a two-night home sleep test (the first day sleeping normally; the second day sleeping with their mouths being taped); their oximetry desaturation index (ODI) and mouth puffing signals (non-mouth puffing, complete mouth puffing, intermittent mouth puffing (IMP), and side mouth puffing) were detected by a validated fingertip pulse oximeter and a mouth puffing detector. The participants were grouped into the ODI-improved group and the ODI-not-improved group according to their sleeping test results. The radiograph was taken by cone-beam computed tomography and cephalometries. Upper airway features including airways, soft tissues, and oral cavity variables were measured.
UNASSIGNED: Participants with severe OSA showed a higher IMP percentage compared with those with normal, mild, and moderate OSA (severe: 33.78%, moderate: 22.38%, mild: 14.55%, normal: 0.31%, p < 0.001). In all participants, the ODI and the percentage of SpO2 under 90 (T90) were positively related to body mass index (BMI) (r = 0.310 and 0.333, respectively), while ODI and T90 were negatively correlated with the minimum width of the airway (r = -0.473 and -0.474, respectively); all mentioned relationships were significant (p < 0.05).
UNASSIGNED: IMP proportions were found to be higher in the half of participants whose ODI did not improve after mouth-taping and in those with severe OSA. Moreover, OSA patients with higher ODI, higher T90, and higher proportions of IMP were more likely to have a narrower upper airway.

An accurate SpO2 value is critical in order to optimally titrate oxygen delivery to patients and to follow oxygenation guidelines. Limited prospective data exist on real-world performance of pulse oximeters in critically ill patients. The objective of this study was to assess accuracy and bias of the SpO2 values measured by several oximeters in hospitalized subjects.
We included stable adults in the ICU with an arterial catheter in place. Main exclusion criteria were poor SpO2 signal and SpO2 > 96%. In each subject, we simultaneously evaluated 4 oximeters: Nonin (Plymouth, Minnesota) embedded in the FreeO2 device (OxyNov, Québec City, Québec, Canada), Masimo (Radical-7, Masimo, Irvine, California), Philips (FAST, Philips, Amsterdam, the Netherlands), and Nellcor (N-600, Medtronic, Minneapolis, Minnesota). Arterial blood gases were drawn and simultaneously each oximeters\' SpO2 values were collected. SpO2 values were compared to the reference (arterial oxygen saturation [SaO2 ] value) to determine bias and accuracy. The ability for oximeters to detect hypoxemia and the impact of oximeters on oxygen titration were evaluated.
We included 193 subjects (153 male, mean age 66 y) in whom 211 sets of measurements were performed. The skin pigmentation evaluated by Fitzpatrick scale showed 96.2% of subjects were light skin (types 1 and 2). One oximeter overestimated SaO2 (Philips, +0.9%), whereas the 3 others underestimated SaO2 (Nonin -3.1%, Nellcor -0.3%, Masimo -0.2%). SaO2 was underestimated with Nonin oximeter in 91.3% of the cases, whereas it was overestimated in 55.2% of the cases with Philips oximeter. Moderate hypoxemia (SaO2 86-90% or PaO2 55-60 mm Hg) was detected in 92, 33, 42, and 11% of the cases with Nonin, Nellcor, Masimo, and Philips, respectively.
We found significant bias and moderate accuracy between the tested oximeters and the arterial blood gases in the studied population. These discrepancies may have important clinical impact on the detection of hypoxemia and management of oxygen therapy.

BACKGROUND: In the diagnosis and primary health care of an individual, estimation of the pulse rate and blood oxygen saturation (SpO2) is critical. The pulse rate and SpO2 are determined by methods including photoplethysmography (iPPG), light spectroscopy, and pulse oximetry. These devices need to be compact, non-contact, and noninvasive for real-time health monitoring. Reflection-based iPPG is becoming popular as it allows non-contact estimation of the heart rate and SpO2. Most iPPG methods capture temporal data and form complex computations, and thus real-time measurements and spatial visualization are difficult.
METHODS: In this research work, reflective mode polarized imaging-based iPPG is proposed. For polarization imaging, a custom image sensor with wire grid polarizers on each pixel is designed. Each pixel has a wire grid of varying transmission axes, allowing phase detection of the incoming light. The phase information of the backscattered light from the fingertips of 12 healthy volunteers was recorded in both the resting as well as the excited states. These data were then processed using MATLAB 2021b software.
RESULTS: The phase information provides quantitative information on the reflection from the superficial and deep layers of skin. The ratio of deep to superficial layer backscattered phase information is shown to be directly correlated and linearly increasing with an increase in the SpO2 and heart rate.
CONCLUSIONS: The phase-based measurements help to monitor the changes in the resting and excited state heart rate and SpO2 in real time. Furthermore, the use of the ratio of phase information helps to make the measurements independent of the individual skin traits and thus increases the accuracy of the measurements. The proposed iPPG works in ambient light, relaxing the instrumentation requirement and helping the system to be compact and portable.

Background Pulse oximeters measure oxygen saturation, heart rate, and perfusion index (PI) by analyzing photoplethysmographic signals. PI is an indirect measure of peripheral perfusion expressed as a percentage of pulsatile signals to non-pulsatile signals. PI measured from different sites may show variation. PI may vary when measured on different fingers. In this study, we aimed to observe the variation of PI among different fingers of both hands. Methodology This cross-sectional, observational study was conducted using a convenience sample recruited from a tertiary care hospital in eastern India. PI was measured in apparently healthy adults in a sitting posture after a five-minute rest. The pulse oximeter probe was attached to each finger and readings were taken after one minute. The analysis of variance and intraclass correlation coefficient (ICC) were calculated to compare and find agreement among PI. Results Data from a total of 391 (229 [58.57%] male and 162 [41.43%] female) adult research participants with a mean age of 34.88 ± 10.65 years were analyzed. The PI was the highest on the middle finger in both hands. There was a significant difference among the PI measured on different fingers, F (9, 3900) = 15.49, p <0.0001. The ICC was 0.474, 0.368, and 0.635 for overall, right-hand, and left-hand fingers, respectively, which indicate poor (ICC < 0.5) to moderate (ICC = 0.5-0.75) levels of reliability. Conclusions The PI measured using consumer-grade pulse oximeters on different fingers may provide different readings. The highest PI reading is found on the middle finger. Clinicians and primary care physicians should consider the differences in measured PI among different fingers and should use the readings with caution for any diagnostic purposes.

Divers who wish to prolong their time underwater while carrying less equipment often use devices called rebreathers, which recycle the gas expired after each breath instead of discarding it as bubbles. However, rebreathers\' need to replace oxygen used by breathing creates a failure mechanism that can and frequently does lead to hypoxia, loss of consciousness, and death. The purpose of this study was to determine whether a pulse oximeter could provide a useful amount of warning time to a diver with a rebreather after failure of the oxygen addition mechanism. Twenty-eight volunteer human subjects breathed on a mixed-gas rebreather in which the oxygen addition system had been disabled. The subjects were immersed in water in four separate environmental scenarios, including cold and warm water, and monitored using pulse oximeters placed at multiple locations. Pulse oximeters placed on the forehead and clipped on the nasal ala provided a mean of 32 s (±10 s SD) of warning time to divers with falling oxygen levels, prior to risk of loss of consciousness. These devices, if configured for underwater use, could provide a practical and inexpensive alarm system to warn of impending loss of consciousness in a manner that is redundant to the rebreather.