How Accurate Is the Owlet Sock? What Studies Show

In FDA-reviewed and published testing, the current Owlet Dream Sock's oxygen saturation (SpO2) readings differ from reference measurements by about 2 to 3 percentage points (root-mean-square error), consistent with its labeled specification, and its pulse rate tracks hospital ECG closely. Accuracy is weaker for alerts: in a 2025 hospital study of 66 infants, the sock's smoothed output rarely alarmed falsely, but it flagged only 14 percent of oxygen drops below 80 percent and 6 percent of heart rate drops below 50 bpm that lasted at least 3 seconds, performing better for longer and milder events. It is an FDA-authorized over-the-counter notification device for healthy infants, not a medical monitor, and there is no evidence that it prevents SIDS.

Product History and FDA Status

Owlet's regulatory status changed in 2021 and 2023. The dates below come from FDA's own records:

• Owlet Smart Sock family (Smart Sock versions 1, 2 and 3, and Smart Sock Plus): Sold with heart rate and oxygen notifications but without FDA marketing authorization. FDA's warning letter to Owlet (dated October 1, 2021, corrected October 5, 2021) stated that products measuring SpO2 and pulse rate to identify desaturation and bradycardia and alarm when readings leave preset values are medical devices, and that the Smart Socks were being marketed without clearance or approval. FDA issued a close-out letter on June 5, 2024, stating that Owlet appeared to have addressed the violations.
• Owlet BabySat 3: A prescription-use pulse oximeter for home spot-checking or continuous monitoring of well-perfused infants older than 1 month and up to 18 months (6 to 30 pounds). FDA cleared it through 510(k) K222597 on June 16, 2023.
• Owlet Dream Sock: FDA granted Owlet's De Novo request (DEN220091) on November 8, 2023. The order created a new device type, "infant pulse rate and oxygen saturation monitor for over-the-counter use" (21 CFR 870.2705, Class II, product code QYU), for healthy infants 1 to 18 months old weighing 6 to 30 pounds.

A De Novo grant is not the same as a 510(k) clearance. De Novo is the route for a novel device type with no legally marketed predicate: FDA makes a risk-based classification and sets special controls that later devices of the same type must meet, including clinical testing of confounding variables such as skin pigmentation. FDA concluded that, with those controls, the Dream Sock's probable benefits outweigh its probable risks for the stated indication. That is not evidence that the device improves infant outcomes.

The authorized indication is narrow. The Dream Sock notifies caregivers when pulse rate or SpO2 moves outside a fixed preset range, so they can assess the infant. FDA's order says it is "not intended to provide notification for every episode" of abnormal pulse rate or low SpO2, only when sufficient data are available, and it is not intended to replace traditional monitoring, diagnosis or treatment.

How the Owlet Measures SpO2 and Heart Rate

According to Owlet's US user manual, the Dream Sock sensor sits in a fabric sock with a flexible strip that curves around the infant's foot, placing the light emitters opposite the detector so that LED light shines through the foot. Peak wavelengths are specified as 649 to 657 nm (red) and 924 to 937 nm (infrared). The ratio of pulsatile red and infrared absorption is used to estimate SpO2, as in other pulse oximeters.

For a detailed explanation of pulse oximetry principles, see our guide on how pulse oximeters work.

FDA's De Novo summary describes the processing:

• Analysis window: The sock computes pulse rate and SpO2 every second from a 10-second rolling window of PPG data, and the app displays a new value every 5 seconds. History graphs in the app show 10-minute averages.
• Motion gating: An accelerometer feeds a signal-quality algorithm. If there is too much motion, the sock outputs no measurement rather than a possibly corrupted one.
• Fixed alarm thresholds: High-priority notifications fire when pulse rate is below 50 bpm or above 220 bpm, or SpO2 is below 80 percent. Owlet's manual states these thresholds are not adjustable.

Accuracy Data

Accuracy here means two different things: how closely individual readings match a reference, and how reliably the alarm detects real events. The Dream Sock looks better on the first than on the second.

Manufacturer and FDA-Reviewed Testing

Owlet's labeling states that SpO2 is accurate to within plus or minus 3 percent Arms over 70 to 100 percent, and pulse rate to within plus or minus 3 bpm over 30 to 300 bpm, both under non-motion conditions (the pulse rate figure was established with a patient simulator). Arms is a root-mean-square error: Owlet's manual notes that only about two-thirds of readings are expected to fall within plus or minus Arms of a co-oximeter value, and that a reading of 90 percent may represent an arterial saturation of 87 to 93 percent.

The studies summarized in Owlet's manual and FDA's decision summary are manufacturer data reviewed by FDA, not independent publications:

FDA's public summary of the same adult study lists slightly different SpO2 values (2.51 and 2.46 percent) and does not legibly label which column is motion. Both versions are below the 3 percent acceptance criterion.

Two details matter. The 70 to 100 percent claim rests on adults, because, as FDA's summary notes, inducing low oxygen in infants would be unethical. And in the 70 to 80 percent band, the band that contains the 80 percent alarm threshold, the manual reports Arms of 3.05 percent without motion and 3.11 percent with motion, slightly above the overall 3 percent specification.

Peer-Reviewed Infant Studies of the Current Generation

Travers et al. (2025). This investigator-initiated study at the University of Alabama at Birmingham was funded by Owlet; the authors state that the sponsor had no role in study design, data interpretation or writing. Its registration (NCT05774470) and full text name the device as the Owlet OSS 3.0, which the authors say received De Novo authorization after testing. Sixty-six hospitalized infants (median 35 weeks' postmenstrual age, at least 1,500 g and off respiratory support) wore the sock for 48 hours alongside hospital ECG and pulse oximetry; the sock's display and alarms were hidden. The sample size matches the 66-patient NICU alarm study described in FDA's De Novo summary.

• Readings: Pulse rate correlated strongly with ECG heart rate (R² 0.96; Arms 2.67 bpm). SpO2 correlated only moderately with the hospital oximeter (R² 0.48; Arms 2.49%), with increasing bias and lower precision when hospital SpO2 was below 80 percent.
• Detection of 3-second events: With the smoothed output used for notifications, sensitivity was 6 percent for heart rate below 50 bpm and 14 percent for SpO2 below 80 percent, with specificity above 99 percent. The raw signal detected more events (39 and 74 percent) but with lower SpO2 specificity (96 percent).
• Longer or milder events: Sensitivity rose with duration and threshold. For example, smoothed sensitivity was 83 percent for SpO2 below 90 percent lasting 10 seconds or more.
• Skin tone: Arms did not differ by Fitzpatrick skin type (I to VI, including 8 infants with types V and VI) or by race, but SpO2 was not compared with arterial blood gases.

The study population was outside the Dream Sock's labeled use. These were preterm and low-birth-weight infants under 44 weeks' postmenstrual age in a NICU, with frequent brief events, whereas the sock is indicated for healthy infants aged 1 to 18 months at home. The authors note that the algorithm was designed for lower-risk infants at home, where false alarms may not be acceptable.

Thomas et al. (2023, print 2024). This study compared the Owlet Smart Sock 3 with a wired Masimo SET oximeter for 60 minutes in 28 preterm or low-weight infants (1.7 to 2.5 kg), 26 of whom were analyzed. Heart rate bias was -1.3 bpm (limits of agreement -6.3 to 3.4); SpO2 bias was 0.3 percent (limits -4.6 to 4.5), with Arms of 2.3 percent over 70 to 100 percent and lower precision at lower SpO2. The paper names the Smart Sock 3 rather than the Dream Sock and does not say which software version was used.

The Earlier JAMA Study (Bonafide et al., 2018)

This research letter tested an older product, before FDA authorization. Bonafide and colleagues compared two consumer monitors, the Owlet Smart Sock 2 and the Baby Vida, against a Masimo Radical-7 reference on the other foot in 30 infants aged 0 to 6 months who were hospitalized on general pediatrics and cardiology wards at the Children's Hospital of Philadelphia in 2017. Each consumer monitor was worn for 60 minutes. Hypoxemia was defined as SpO2 below 91 percent and bradycardia as pulse rate below 90 bpm. The figures below are for the Owlet unless stated otherwise:

• Hypoxemia: Sensitivity was 88.8 percent (95% CI 79.4 to 98.1) and specificity 85.7 percent (95% CI 72.6 to 98.8), based on 262 sampled readings from 28 infants. All 12 infants who had hypoxemia on the reference monitor had at least one simultaneous low reading on the Owlet, but 5 of the 12 also had at least one stable normal reading during hypoxemia.
• Bradycardia: The Owlet never displayed a falsely low pulse rate. True bradycardia was too rare (0.8 percent of sampled readings) to estimate its sensitivity meaningfully.
• Baby Vida, for contrast: It detected none of the hypoxemia (sensitivity 0 percent), and in 14 infants it displayed falsely low pulse rates.

The authors concluded that the Owlet "detected hypoxemia but performed inconsistently". The study used a pulse oximeter rather than arterial blood gases as the reference, and it does not describe the current Dream Sock.