Tidepool Big Data Donation Project

Participating in Tidepool Big Data Donation Project is opt-in, all users must choose to participate. All donated data is anonymized.

Tidepool operates the Tidepool Big Data Donation Project, which gives Tidepool users the opportunity to donate their anonymized data to support diabetes research and innovation.

Tidepool may license donated datasets to third parties for a fee.

When a clinician user creates a new account for a patient, that is considered a Custodial account. When an email address for the patient is added to that Custodial account, an email is automatically delivered to the patient giving them the opportunity to claim their account.

If a patient completes the process to claim their account, the patient now assumes full control of the account. Then, and only then, can the patient choose to participate in Tidepool Big Data Donation Project. The clinic will still have access to see data in that account, as if the patient had created the account themselves at home and shared their data with the clinic.

If the clinic does NOT enter a patient email address for a Custodial account, the account will remain under the control of the clinic, and the patient will not have access to the data from home, nor be able to upload their own data from home.

Custodial accounts created by a clinic are never presented with the choice to opt-in to participate in Tidepool Big Data Donation Project, as such, no data uploaded to a Custodial account is made available in an anonymized form to third parties through Tidepool Big Data Donation Project.

We anonymize data in these ways:

For each user:
- Included: The month and year of birth and diagnosis
- Not included: Name, address, email address, exact birthday, exact diagnosis day, notes, photos, customized basal profile names or other personally identifiable information
For all diabetes devices:
- Included: Device event and data timestamps
- May be included: Brand and model of the device (some device makers preclude this)
- Not included: Device serial number
For blood glucose monitors:
- Included: Blood glucose (BG) readings
For continuous glucose meters (CGM):
- Included: Estimated glucose values, events tracked by the CGM, including meals, insulin, calibration, exercise
For insulin pumps:
- Included: All pump settings, including bolus calculator parameters, basal rates, insulin to carb ratios, and insulin sensitivity factors as well as all events tracked by the pump, including meals, insulin dosed, temp basals and suspends events, primes, and BG inputs

Note: Some of the terms above may have slightly varying language on your device.

We will not include in the anonymized datasets (1) freeform text and notes entered by you or your Care Team Members, or (2) any other data that could identify a specific individual.

See our Privacy Policy for more information.

After anonymizing and removing all identifying information, the data we provide researchers is in either .csv, .json, or .xlsx format that looks like this.

We store data in Tidepool’s secure, HIPAA-compliant servers that are hosted by Amazon Web Services (AWS). Within AWS, we use MongoDB, a popular, modern database. See System Architecture for additional details about data is secured.

Here is an example data file (6MB compressed) for our demo user, “Jill Jellyfish" before anonymization. It follows the Tidepool data model (the way we internally represent data) here.