Data Repository

The Open Oximetry Project is working to accelerate improvements in device performance through data sharing.

About Our Data


Initially, the majority of data in the repository has been collected by the OpenOximetry Team based at the UCSF Hypoxia Lab. These data are de-identified and collected from both healthy human adult volunteers in controlled desaturation studies, as well as critically ill patients in the intensive care unit. Studies are conducted in accordance with IRB approved protocols and study subject consent. The data repo build began in October 2023, so please be patient with us as we work to incorporate data.

We invite the global community to contribute data to this repository so long as it adheres to our data dictionary format, data collection protocols and data sharing agreements.

Repository Purpose

We hope that by creating and sharing this unique data set, we will be able to collectively accelerate improvement of pulse oximeter performance and reduce health disparities.

Use Cases

The Open Oximetry team is collecting data that we want to share with the global community of stakeholders.

Researchers : : Product Developers : : Consumers & Patients : : Clinicians : : Policy & Procurement Officials : :  Regulatory Agencies

Data Terms of Use

Downloading Data

All data collected by the Open Oximetry project are available (de-identified) as open access in our repository. We also include data from collaborators who are collecting data using protocols consistent with Open Oximetry data protocols and standards. To download data, use the link above. You can view the full Terms of Use for our site here and the Terms of Use for the Data Repository here. We also will ask users to comply the data host’s terms of use (

Uploading Data

We invite data from collaborators who are collecting data using protocols consistent with Open Oximetry data protocols and standards. The full list of data collection protocols can be found on our Protocols page and the Data Dictionary can be found using the link above. All data uploaded must be done in accordance with our data safety protocols, local IRB approvals and other relevant regulations, and be allowed to be shared in accordance with the Terms of Use for the Open Oximetry Repository.

Publishing Analyses

Users are encouraged to publish analyses of data from the repository. Such publications should acknowledge and cite the repository as: “Open Oximetry Data Repository. The Open Oximetry Project. 2023. Accessed at”

Read Terms of Use


Yes, data are de-identified. Participants are given a unique participant ID so that they can be identified between sessions (i.e. each participant may have many sessions). No chronology of patients can be inferred by attempting to reverse-engineer the patient IDs. Dates in the session table are randomly offset into the future by a random offset (set per participant) while attempting to preserve seasonality. This means that, for example, it would not be possible to ascertain whether one participant had a session before another participant (i.e. it would not be possible to ascertain inter-participant chronology), however, it would be possible to compute the amount of time between sessions for a singular participant who enrolled in multiple sessions (i.e. it would be possible to ascertain intra-participant chronology).

Data are stored either as flat CSV files (for tabular data) or in WFDB format. See the Data Dictionary for more information.

The OpenOximetry database consists of pulse oximetry data composed of laboratory testing data from participants at the Hypoxia Lab, at the University of San Francisco California (and soon from collaborating labs that adhere to shared data collection principles for these types of studies). Study participants are placed on a closed breathing circuit, allowing study investigators to intentionally desaturate participants in a controlled manner by titrating oxygen and nitrogen levels in the circuit. Data captured during these studies includes paired pulse oximeter (SpO2) and arterial blood gas readings (PaO2) throughout a range of oxygen saturations, patient demographics, longitudinal quantitative and qualitative skin color measurement (using the Monk Scale and spectrophotometry measurements at multiple anatomic locations), continuous ECG, arterial blood pressureBP, and EtCO2/EtO2, and unprocessed PPG. Additionally, data from point-of-care arterial blood gas monitoring (SaO2) as well as pulse oximeter (SpO2) readings are captured during study sessions.

Yes. Information and data standards for posting your data to the repository will be coming soon.

Submit a question