There are four main ways to get continuous glucose monitor (CGM) data into your software: a manufacturer’s developer API, such as the Dexcom API; a formal partner integration with the CGM maker; the phone’s health platform, Apple HealthKit or Android Health Connect; or a third-party data platform that has already done the integrations. Each route differs in which sensors it covers, how fast the data arrives, and what approvals you need. For the clinic, CGM data can also flow into the electronic health record (EHR) through Epic integrations and the new HL7 FHIR CGM standard.
This guide is for teams building diabetes apps, remote monitoring platforms, clinical study software, and wellness products. It compares the routes, explains what each one delivers, and covers the regulatory and privacy questions that come with handling glucose data.
The four routes at a glance
Route | Examples | Data speed | Best for |
|---|---|---|---|
Manufacturer developer API | Dexcom API (v3) | Dexcom: 1-hour delay in the US, 3 hours elsewhere | Apps that need historical CGM data from Dexcom users, with the user’s consent |
Partner integration | Dexcom real-time partner APIs, Abbott FreeStyle Libre partnerships | Near real time, depending on the agreement | Insulin delivery, connected pens, and commercial platforms with a business relationship |
Phone health platform | Apple HealthKit, Android Health Connect | Dexcom G7 and Stelo write to Apple Health with a 3-hour delay | Wellness and lifestyle apps that read data the user already shares |
Data platform or aggregator | Tidepool, Glooko, Validic | Depends on the platform and its source | Apps that need several devices through one integration, or EHR delivery |
Which route fits depends on three questions. Which sensors do your users wear? How fresh does the data need to be? And will your software only show and store the data, or analyze it to drive alerts or treatment?
The Dexcom API
Dexcom runs the most open developer program among the major CGM makers. Its developer API, now at version 3, gives apps access to a user’s data once that user signs in and authorizes the app.
What the API provides. Version 3 endpoints return:
- Estimated glucose values (EGVs), the core CGM readings
- Events the user logged, such as meals, insulin, and exercise
- Alerts, calibrations, and device information
- The date range of available data
The API supports data from the Dexcom G6, G7, and G7 15 Day. Dexcom’s documentation does not list Stelo as a supported device.
How fast the data arrives. Dexcom’s endpoint documentation states that data is available “with a one-hour delay in the United States” and “with a three-hour delay outside the United States.” That makes the API right for trends, reports, coaching, and research, and wrong for anything that needs to react to a glucose reading in real time.
How access works. Every developer starts in a free sandbox with test data. Production use requires OAuth 2.0 user authentication and HIPAA authorization from each user, and Dexcom sets three access levels:
Level | What it allows |
|---|---|
Registered | Sandbox access with test users |
Limited | Real data for up to 5 authorized users, after review by Dexcom’s partnerships team |
Full | Commercial use, after Dexcom’s technical and commercial review |
Requests are limited to 60,000 calls per app per hour. Plan the Full access review into your timeline: it determines when you can launch.
Real-time data. Real-time access is a separate partner route. In July 2021, Dexcom announced FDA clearance of its Partner Web APIs and invited third parties to access real-time data, naming Garmin and Livongo as early partners. Real-time access comes through a partnership with Dexcom, not the standard developer program, so ask Dexcom early about current partner terms.
Abbott FreeStyle Libre and other manufacturers
Abbott does not offer a public developer API for FreeStyle Libre data. Instead, it works through named partnerships, including integrations with insulin delivery systems from MiniMed, Insulet, Tandem, Beta Bionics, and Sequel, connected pens from Novo Nordisk, and digital health companies such as Omada Health.
Unofficial libraries that read LibreView data exist, but Abbott does not support them. They are a poor foundation for a commercial or regulated product, because an unannounced change can break them, and they sit outside any agreement covering data use.
Medtronic also does not offer a public third-party API for CareLink data. Other platforms have accessed its data through agreements, such as its 2015 agreement with Glooko.
For products that need Libre or Medtronic data, the practical routes are partnering with the manufacturer, using a data platform that already has one, or writing to the phone’s health platform where the manufacturer’s app already writes.
Apple HealthKit and Android Health Connect
Phone health platforms are the simplest route for consumer apps: if the CGM maker’s app writes glucose data to the platform, and the user grants permission, any app can read it.
- Apple HealthKit stores glucose as HKQuantityTypeIdentifierBloodGlucose, in mg/dL or mmol/L depending on region.
- Android Health Connect stores glucose as a BloodGlucoseRecord and requires READ_BLOOD_GLUCOSE and WRITE_BLOOD_GLUCOSE permissions.
The tradeoff is delay. Dexcom says Apple Health receives Dexcom G7 data “after a three-hour delay, not in real time,” and Stelo sends data to health apps with the same 3-hour delay. That works for daily summaries, meal insights, and wellness trends. It does not work for alerts.
Partnerships can go further. Dexcom’s Stelo syncs with the Oura ring app, adding glucose to Oura’s sleep and activity data, following a strategic partnership the two companies announced in November 2024.
Data platforms and aggregators
Data platforms handle the device integrations so that your app integrates once.
- Tidepool offers an HTTP REST API and open-source software. Developers request a client identifier for their application.
- Glooko collects data from many diabetes devices, including Dexcom and Medtronic systems, for patients and clinics.
- Validic announced in November 2024 that it delivers CGM data into Epic and Oracle Health EHRs, with support for Abbott’s FreeStyle Libre portfolio and other CGMs.
A platform saves integration work, but it adds a dependency. Check which devices it covers, how fresh the data is for each, and what its contract says about data ownership and use.
Getting CGM data into the EHR
Clinicians want CGM data where they already work. Three developments make that more practical:
- Dexcom’s Epic integration. Dexcom’s direct EHR integration automatically pushes data from G7 app users into Epic, with AGP and GMI reporting. It requires Epic version May 2023 or newer and MyChart.
- Abbott’s Epic integration. In April 2025, Abbott announced that LibreView data can be automatically linked to Epic in the US through Epic’s Aura software.
- The HL7 FHIR CGM Implementation Guide. Published as STU 1 (version 1.0.0) in September 2025, it defines a standard way to send CGM data to an EHR on FHIR R4. It covers summary metrics such as mean glucose, time in range, GMI, coefficient of variation, and sensor wear, along with device information and individual readings. A submission can include any combination, so raw readings are not required.
For developers, the FHIR CGM guide is the target. It works with SMART on FHIR, so one integration can serve any EHR that adopts the standard, instead of one custom interface per health system. The Diabetes Technology Society’s iCoDE project, launched in 2022, has also worked on standards for bringing CGM data into health records. See our guides to diabetes device interoperability and why EHR interoperability is critical.
Regulatory and privacy questions
Showing data vs analyzing it
What your software does with CGM data decides whether it is a medical device.
- Transfer, storage, and display are generally not devices. FDA’s Medical Device Data Systems guidance says software functions “solely intended to transfer, store, convert formats, or display medical device data and results” are not devices.
- Analysis and alerts are devices. The same guidance excludes software intended “to interpret or analyze” device data, functions for active patient monitoring, and software that generates alarms or alerts. FDA also lists a product code for CGM secondary displays (PJT), which is Class II and exempt from 510(k).
- Clinician decision support has limits. Under FDA’s 2026 guidance, CGM readings are a pattern or signal, so software that analyzes them does not qualify as non-device clinical decision support. See our guide to clinical decision support software.
Plan the regulatory assessment before you design features. A dashboard that shows glucose trends and a feature that flags a coming low can sit next to each other in the same app and fall on opposite sides of the line. For how the FDA treats prediction, dosing, and AI features, see our guide to AI in diabetes devices.
HIPAA or the FTC
Who must protect the data depends on your relationship with the user.
- Consumer apps chosen by the user. HHS says that when a patient sends data to an app they chose, not provided by or for a covered entity, the information “is no longer subject to the protections of the HIPAA Rules.” In HHS’s health app scenarios, a consumer app is not a business associate.
- Apps provided for a provider or health plan. When a covered entity contracts with you to provide the app, you are typically a business associate, and HIPAA applies. See our guide to HIPAA in MedTech development.
- The FTC fills the gap. The FTC’s amended Health Breach Notification Rule, effective July 29, 2024, covers many health apps that HIPAA does not, and requires notice to users and the FTC after a breach.
Dexcom’s API itself requires HIPAA authorization from each user before you receive production data, so build consent into onboarding from the start.
Architecture checklist for CGM data
- Map sources to sensors. List which CGMs your users wear and which route covers each one.
- Design for the delay. Build features around the data speed you will actually get, and label data age clearly in the interface.
- Normalize units and time. Store mg/dL or mmol/L consistently, handle time zones, and keep source timestamps.
- Keep provenance. Record where each reading came from and when it arrived, so you can trace and correct data later.
- Handle gaps. CGM data has warm-up periods, signal losses, and sensor changes. Show gaps honestly, and do not fill them with estimates presented as readings.
- Separate display from analysis. Keep regulated functions, such as alerts and analysis, in their own components, so the rest of the product can change without new regulatory work.
- Build consent and security in. Manage OAuth tokens securely, log access, and support data deletion requests.
- Plan for the EHR. If clinicians are part of your product, design your data model around the FHIR CGM guide’s summary metrics.
Frequently asked questions
Does Dexcom have an API?
Yes. The Dexcom Developer API (version 3) gives apps access to a user’s CGM data after the user authorizes the app. It returns glucose readings, events, alerts, calibrations, and device information for the Dexcom G6, G7, and G7 15 Day. In the US, data arrives with a one-hour delay. Commercial use requires Dexcom’s Full Access review.
Is there a FreeStyle Libre API?
Abbott does not offer a public developer API for FreeStyle Libre. Libre data reaches other products through Abbott’s named partnerships, through data platforms with Abbott agreements, and through Abbott’s Epic integration. Unofficial libraries exist but are not supported by Abbott.
Can I get real-time CGM data from Dexcom?
Not through the standard developer API, which delivers US data with a one-hour delay. Dexcom has offered real-time data to partners through its Partner Web APIs, which were FDA-cleared in 2021, so real-time access requires a partnership with Dexcom.
How does CGM data get into Epic?
Dexcom’s direct EHR integration automatically sends data from G7 app users to Epic, and Abbott links LibreView data to Epic through Epic’s Aura software. Data platforms such as Validic also deliver CGM data into EHRs. The HL7 FHIR CGM Implementation Guide, published in 2025, sets a standard format for sending CGM data to any EHR that adopts it.
Is software that displays CGM data a medical device?
Software that only transfers, stores, converts, or displays CGM data is generally not a device under FDA’s Medical Device Data Systems guidance. Software that analyzes the data, generates alarms or alerts, or supports active patient monitoring functions as a device.
Build CGM data integrations with Sequenex
CGM integration looks simple until you hit the details: access reviews, data delays, units and time zones, consent, regulatory boundaries, and EHR delivery.
Sequenex builds the mobile and cloud software for biosensor, CGM, and connected diabetes products, including device connectivity, data platforms, clinician dashboards, and EHR integration. See our biosensor and CGM software development services, or contact us to discuss your integration.
