BDMOXMTRBLK

We categorize devices as fingertip, handheld, tabletop, multiparameter, phone-based or wearable.

Some devices may have the capability to function with transmission or reflectance probes. Read more about the difference between reflectance and transmission devices in our FAQ.

This indicates the intended patient populations for the device (adult, pediatrics, neonates), as specified by our review of the manufacturers' published specifications. Use in certain patient populations may require procurement of a separate probe.

This indicates the location where the device is manufactured as stated by the manufacturer (or the stated location of the manufacturer). Please note, devices may contain components manufactured in different location.

"Ingress Protection" ratings define levels of sealing effectiveness of devices from foreign bodies (e.g. dust) and moisture. Read more at our FAQ.

Here we report retail purchase costs (USD) for buying the pulse oximeter, including one adult finger probe. Costs are obtained from one or multiple sources including manufacturers or online retail stores. Of note, some devices have special discount pricing for low and middle-income countries. The special prices are not accounted for in this report.

*CAUTION: This calculator makes many assumptions, is not based on the specific manufacturer/model durability, and costs may not reflect cost of ownership in well-resourced or home care settings. Costs are based solely on the purchase cost for this oximeter and general assumptions about how long a device with this form factor (i.e., fingertip vs handheld vs tabletop) might be expected to last in a heavy-use environment (e.g., a resource-variable clinical setting with frequent use, where damage, loss and theft are possible). These assumptions were informed by input from clinicians across diverse settings. Lifetime ownership cost varies considerably by user and setting. Click the settings button next to the cost to view the formula and adjust these assumptions based on your local context and use case.

Here we report device features such as signal quality indicator, waveform, carboxy-Hb, perfusion index and ability to measure Hb. These are based on review of manufacturers' manuals and may be incomplete.

Here we report the root mean square error (ARMS) as provided in the manufacturer’s product manual or other literature, which may include data from the 510(k) submission.

Here we report pulse oximeter performance based on Open Oximetry Project device testing using 2013 FDA Guidelines for 510k clearance and 2017 ISO 80601 standards, while also trying to account for expanded criteria to improve diversity of skin pigment in study cohorts (US FDA "Approach for Improving the Performance Evaluation of Pulse Oximeter Devices Taking Into Consideration Skin Pigmentation, Race and Ethnicity"). You can view our study protocols here, or more generally read about quantification of oximeter performance on our FAQ.

Root mean square error (ARMS) is a common measure of pulse oximeter device performance that combines bias and precision.

*NOTE: performance is only reported here once we have tested the device in ≥10 study subjects (i.e. as required by 2013 FDA and 2017 ISO requirements). Performance may change significantly as we continue to perform testing in additional subjects and conditions. Please continue to check back as we update frequently. Click the device to see details on how many subjects have been tested as well as details of skin color testing.

Currently, there is lack of consensus on optimal study cohort sizes for pulse oximeter validation studies. The 2017 ISO 80601 and 2013 FDA regulatory frameworks stipulate at least 10 subjects, 15% of whom should be darkly pigmented.

We report device results as preliminary once at least 10 subjects have been tested, though continue testing devices in as many diverse participants as we can. We are awaiting updated regulatory guidance for optimal cohort sizes.

Currently, there is lack of consensus on optimal methods for characterizing skin pigment and optimal sample sizes for validation study cohorts. 2017 ISO and 2013 FDA documents stipulate at least 10 subjects, 15% of whom should be darkly pigmented. Here we define ‘dark skin pigmentation’ as Monk Skin Tone Scale HIJ and Individual Typology Angle <-30. For the purpose of data analysis and to avoid operator bias from assigning MST, we use ITA >30 for light, ITA 30 to -30 for medium, ITA < -30 for dark, and ITA < -50 for very dark.

This is the most recent date that the Open Oximetry Project collected data in the UCSF Hypoxia Lab to assess this device's performance. If ARMS data were obtained from a source other than the Hypoxia Lab, please review the date for that source. Of note, device performance may be specific to a model year (even if the model name has not changed).

Root mean square error (ARMS) is a common measure of pulse oximeter device performance. 'ARMS' may be ascertained from manufacturers' published data, 510k reports, package inserts or primary data from testing conducted by the UCSF Hypoxia Lab. Devices independently tested by the Open Oximetry Project will be marked 'verified' or 'failed' depending on study findings.

Here are select manuscripts published in the peer-reviewed literature.

We are working to gather device performance data published directly by manufacturers.

This figure shows the forehead skin color for healthy volunteer participants in the study cohort tested with this pulse oximeter. Each square represents one subject, with the square’s color corresponding to the Monk Skin Tone Scale category observed by UCSF Hypoxia Lab clinical research coordinators. More info on skin color quantification.

All devices were tested in participants with light, medium and dark skin pigment as defined by Individual Typology Angle. The value reported here indicates the percentage of individuals from the 'dark' pigment cohort (i.e. ITA < -30) who also had Individual Typology Angle <-50 (i.e. 'very dark' pigment).

This number (i.e. 'differential bias’ or 'disparate bias') describes how much pulse oximeter performance is impacted by skin pigment. This is done by using real data to model what happens if we compare how accurate SpO2 is (i.e. how SpO2 compares to gold standard blood SaO2 co-oximetry) for a healthy volunteer with very light skin pigment and SpO2 accuracy for a healthy volunteer with very dark skin pigment (i.e., an ITA difference of 100), by subtracting the difference. Read more about differential bias here in our FAQ.

This number (i.e. 'differential bias’ or 'disparate bias') describes how much pulse oximeter performance is impacted by skin pigment. This is done by using real data to model what happens if we compare how accurate SpO2 is (i.e. how SpO2 compares to gold standard blood SaO2 co-oximetry) for a healthy volunteer with very light skin pigment and SpO2 accuracy for a healthy volunteer with very dark skin pigment (i.e., an ITA difference of 100), by subtracting the difference. Read more about differential bias here in our FAQ.

This shows the average difference between the oxygen levels measured by the pulse oximeter (SpO2) and the actual blood oxygen levels (SaO2) for participants with dark skin pigment when their oxygen levels are between 70-84%. Negative values indicate the pulse oximeter SpO2 underestimates actual blood oxygen SaO2. Positive values indicate pulse oximeter SpO2 overestimates actual blood oxygen SaO2.

This shows the average difference between the oxygen levels measured by the pulse oximeter (SpO2) and the actual blood oxygen levels (SaO2) on average for participants with medium skin pigment when their oxygen levels are between 70-84%. Negative values indicate the pulse oximeter SpO2 underestimates actual blood oxygen SaO2. Positive values indicate pulse oximeter SpO2 overestimates actual blood oxygen SaO2.

This shows the average difference between the oxygen levels measured by the pulse oximeter (SpO2) and the actual blood oxygen levels (SaO2) on average for participants with light skin pigment when their oxygen levels are between 70-84%. Negative values indicate the pulse oximeter SpO2 underestimates actual blood oxygen SaO2. Positive values indicate pulse oximeter SpO2 overestimates actual blood oxygen SaO2.

This shows the average difference between the oxygen levels measured by the pulse oximeter (SpO2) and the actual blood oxygen levels (SaO2) on average for participants with dark skin pigment when their oxygen levels are between 85-100%. Negative values indicate the pulse oximeter SpO2 underestimates actual blood oxygen SaO2. Positive values indicate pulse oximeter SpO2 overestimates actual blood oxygen SaO2.

This shows the average difference between the oxygen levels measured by the pulse oximeter (SpO2) and the actual blood oxygen levels (SaO2) on average for participants with medium skin pigment when their oxygen levels are between 85-100%. Negative values indicate the pulse oximeter SpO2 underestimates actual blood oxygen SaO2. Positive values indicate pulse oximeter SpO2 overestimates actual blood oxygen SaO2.

This shows the average difference between the oxygen levels measured by the pulse oximeter (SpO2) and the actual blood oxygen levels (SaO2) on average for participants with light skin pigment when their oxygen levels are between 85-100%. Negative values indicate the pulse oximeter SpO2 underestimates actual blood oxygen SaO2. Positive values indicate pulse oximeter SpO2 overestimates actual blood oxygen SaO2.

We collected the percent modulation of the infrared (IR) signal—often termed pulsatility amplitude or perfusion index (PI)—from our reference devices, Nellcor PM1000N and Masimo Rad-97. This measure serves as an indirect, albeit imperfect, surrogate for perfusion and signal strength.

This column provides insights into the pulsatility amplitude distribution of the cohort by reporting the median pulsatility amplitude and interquartile range (Q1–Q3). The median pulsatility amplitude represents the typical blood flow strength of the collected samples, while the interquartile range (Q1-Q3) shows the middle 50% of pulsatility amplitude values, giving an idea of how much variation exists around this typical flow level.

We report pulsatility amplitude from the Nellcor PM1000N for all devices, except for the Masimo Rad-97, which provides its own PI measurement. For consistency across devices, we applied a correction factor to the PM1000N pulsatility amplitude by dividing the values by 10, making them comparable to the pulsatility amplitude reported by other devices, such as Masimo.

We measured the percent modulation of the infrared (IR) signal, also referred to as pulsatility amplitude or perfusion index (PI). This measurement is sometimes used as an imperfect and indirect surrogate for perfusion and signal strength.

Here, we show the percentage of the cohort tested with this device who had low pulsatility amplitude, defined as a pulsatility amplitude value less than 1, based on values reported by one of our reference devices, the Nellcor PM1000N.

To ensure consistency with other reference devices, we divide the pulsatility amplitude from the PM1000N by 10. It is important to note that this correction factor makes the PM1000N values comparable, though not identical, to the PI reported by devices like Masimo.

We are working to gather raw data for device performance to share for independent analysis. We expect to launch this feature soon.

We are working on novel in vitro testing protocols for both commercially available devices (e.g. Fluke ProSim8) and novel in vitro devices. We expect to report data for this testing soon.

Here we link to studies conducted in the clinical settings.