The OpenAPS Open Source Ecosystem and Developer Contributions

OpenAPS—short for Open Artificial Pancreas System—is one of the most influential open source initiatives in the diabetes technology space. Born from the frustration of patients and engineers with proprietary, slow-moving medical devices, the project has grown into a global collaborative effort that redefines what is possible for automated insulin delivery. By combining continuous glucose monitors (CGMs), insulin pumps, and open source software, OpenAPS enables individuals with type 1 diabetes to achieve tighter glucose control with less manual intervention. The ecosystem surrounding OpenAPS is not just a set of scripts and hardware; it is a vibrant community of developers, clinicians, researchers, and users who together push the boundaries of what a patient-driven medical system can achieve.

This article provides an in-depth exploration of the OpenAPS ecosystem, focusing on the technical architecture, the variety of developer contributions, and the broader impact of open source collaboration in healthcare technology. Whether you are a developer interested in contributing or a person with diabetes evaluating open source options, understanding the inner workings of OpenAPS offers valuable insights into the future of patient-centered medical innovation.

A Brief History and Motivation

The OpenAPS project began in 2013 when Dana Lewis and Scott Leibrand set out to build a system that could automate insulin delivery using existing off-the-shelf hardware. At the time, commercial artificial pancreas systems were still in clinical trials, and those that existed were inaccessible, expensive, or locked down. The core idea was to create a "does-it-yourself" approach: write software that runs on a small computer (initially a Raspberry Pi or Intel Edison) that reads CGM data, calculates insulin dosages using a model predictive control or proportional-integral-derivative algorithm, and communicates wirelessly with an insulin pump via a radio bridge like the RileyLink. The first successful closed-loop experiments were documented on Dana’s blog, and the code was released on GitHub under a permissive open source license. That simple act of publishing code sparked a movement.

Since then, the project has evolved through multiple repositories, including openaps, oref0, and oref1, with contributions from hundreds of developers worldwide. The ecosystem now supports a variety of hardware configurations, from the original Raspberry Pi setup to more modern solutions running on Android phones (AndroidAPS) and Apple devices (Loop). The shared philosophy remains: any person with diabetes should have the ability to build, customize, and control their own automated insulin delivery system, provided they are willing to take on the responsibility of understanding the risks.

Technical Architecture of the OpenAPS Ecosystem

The OpenAPS stack can be understood as several layers: hardware, communication bridge, control algorithm, and user interface. Each layer has seen numerous innovations and contributions from the community.

Hardware Layer: CGMs, Pumps, and Bridges

  • Continuous Glucose Monitors (CGMs): OpenAPS typically uses Dexcom G4, G5, G6, or Libre sensors (the latter requiring a transmitter like MiaoMiao or Bubble). The CGM provides near-real-time glucose readings, typically every 5 minutes.
  • Insulin Pumps: Historically, older Medtronic pumps (e.g., 515, 715, 522, 722, 523, 723) were the most common because they use a proprietary RF protocol that could be reverse-engineered. More recent work includes support for Omnipod (via Omnipod DASH, Eros, or Omnipod 5 in some configurations) and older Animas pumps (though Animas discontinued production).
  • RileyLink and Bridges: The RileyLink (originally designed by Pete Schwamb) is a small circuit board that bridges the gap between a pump’s RF signals and a device that can run the OpenAPS software (Raspberry Pi, Intel Edison, or a phone via Bluetooth). It uses a CC1111 radio chip to communicate with the pump and a Bluetooth module to talk to the control device. Later versions like the RileyLink 2.0 and the OrangeLink have improved range and reliability.

Software Layer: Control Algorithms

The brain of OpenAPS is the oref0 (OpenAPS Reference Design 0) and now oref1 algorithm. These are JavaScript/node.js applications that run the core loop:

  1. Read CGM data via a configured source (Dexcom Share API, Nightscout, or local upload).
  2. Calculate a predicted glucose curve using a model of insulin absorption (based on the user’s insulin sensitivity, carbohydrate ratio, and duration of insulin action) and meal announcements.
  3. Determine a temporary basal rate (increase or decrease) to keep glucose within target range, often using a basal-microbolusing logic that re-evaluates every 5 minutes.
  4. Communicate the command to the pump via the RileyLink.

The algorithm includes safety features such as low glucose suspend, high glucose correction thresholds, and optional super micro bolus (SMB) for faster response. All decisions are logged and can be reviewed in Nightscout, an open source data visualization tool that acts as the primary user interface and remote monitoring system.

User Interface and Monitoring: Nightscout

Nightscout (also known as CGM in the Cloud) is a companion open source project that displays CGM data, insulin delivery, and predictions on a web dashboard. Users can share their data with caregivers, clinicians, or the developer community for troubleshooting. Nightscout also provides APIs that OpenAPS uses for data input and output. The tight integration between Nightscout and OpenAPS means that all modifications to the system can be monitored in real time, providing an extra layer of safety.

Developer Contributions: The Engine of the Ecosystem

The OpenAPS project is maintained almost entirely by volunteers—people with type 1 diabetes, parents of children with diabetes, software engineers, and healthcare professionals. Their contributions fall into several key areas:

Code Contributions and Feature Development

The primary repository, OpenAPS/oref0 on GitHub, has seen over 2,000 commits from more than 120 contributors. Major features such as the super micro bolus (SMB), advanced unannounced meal handling, and dynamic ISF (insulin sensitivity factor) were contributed by community developers. The open source workflow follows a standard model:

  • Issue tracking: Users report bugs or request features through GitHub Issues.
  • Pull Requests (PRs): Developers fork the repository, implement changes, and submit PRs. Core maintainers review code for safety, performance, and alignment with the project’s goals.
  • Testing and Beta Channels: The community runs nightly builds and dedicated branches (e.g., dev for bleeding-edge features) to test new algorithms before they are released to the general user base.

Documentation and Education

One of the biggest barriers to entry for new users is the complexity of setting up OpenAPS. The community has created extensive documentation on the OpenAPS website, including a step-by-step “Build Your Own OpenAPS” guide, troubleshooting FAQ, and algorithm explanation diagrams. Developers contribute by:

  • Writing and updating wiki pages on GitHub.
  • Creating video tutorials and blog posts (e.g., Dana Lewis’s original blog).
  • Translating documentation into other languages (e.g., German, Spanish, Dutch).
  • Maintaining the OpenAPS/docs repository.

Integration with Other Open Source Tools

The OpenAPS ecosystem does not operate in a silo. Developers have built bridges to other projects, creating a web of interoperable tools:

  • AndroidAPS: A port of the OpenAPS algorithm to Android phones, enabling a phone-based closed loop.
  • Loop: An iOS-based closed loop system that uses the same core logic but with a different communication stack.
  • Tidepool: A HIPAA-compliant data platform that can pull data from Nightscout for clinical review.
  • xDrip+: A powerful CGM collector and display app for Android that feeds data to OpenAPS.

Each integration project has its own developer community, but many core OpenAPS contributors are active across multiple repositories, sharing knowledge and code.

Hardware Hacking and Reverse Engineering

Without access to official pump communication protocols, the OpenAPS community had to reverse-engineer the wireless commands used by Medtronic pumps and, later, the Omnipod Eros and DASH systems. Developers like Pete Schwamb (RileyLink) and Scott Hanselman (pump code analysis) played pivotal roles. This work involves analyzing RF signal captures, understanding encryption mechanisms (or lack thereof), and writing low-level radio code in C or JavaScript. The result is a set of open libraries that allow any developer to build new devices or interfaces.

Quality Assurance and Safety Review

Because OpenAPS is a medical device system, safety is paramount. Developers contribute by:

  • Automated testing: Running unit tests and integration tests on the algorithm using historical data sets.
  • Code review: Each PR is reviewed by at least two experienced contributors, with special attention to scenarios that could lead to over-delivery or under-delivery of insulin.
  • Real-world testing: Many contributors run multiple systems simultaneously to compare algorithm behavior.
  • Bug bounty programs: Some community members offer rewards for identifying edge cases or security vulnerabilities.

Impact of Open Source Collaboration on Diabetes Technology

The OpenAPS project has had a profound impact beyond its user base. It has pressured traditional medical device manufacturers to accelerate their own development of hybrid closed-loop systems. It has demonstrated that patient-driven innovation can be safe when combined with transparent code, extensive real-world data, and community oversight. Academic institutions have studied OpenAPS outcomes, publishing papers that show improvements in time-in-range and reduction in HbA1c without increased hypoglycemia.

Furthermore, the ecosystem has become a training ground for a new generation of health-tech developers. Many contributors have moved on to work at medtech startups or have launched their own ventures (e.g., Loop’s creator working on a commercial product). The knowledge gained from reverse-engineering pumps has also informed regulatory discussions about data access and interoperability.

Challenges and Limitations

Despite its successes, the OpenAPS ecosystem faces several challenges:

  • Regulatory gray area: In many countries, building a system that modifies insulin delivery could be considered practicing medicine without a license. Users assume all risk, and the community provides disclaimers.
  • Access to hardware: As Medtronic phases out older pumps, users must find used devices on secondary markets. Newer pumps like the Tandem t:slim X2 use proprietary communication that has not been fully opened.
  • Frequency of updates: Staying current with algorithm improvements requires active attention. A user who builds a system and never updates it may miss important safety fixes.
  • Learning curve: The technical knowledge required to build and troubleshoot OpenAPS can be overwhelming for non-developers.

The community continues to address these challenges through better onboarding, automated tools (like the OpenAPS Toolkit), and advocacy for open data access.

Future Directions and How to Get Involved

The OpenAPS ecosystem is far from static. Current development areas include:

  • Algorithm improvements: Using machine learning to predict glucose excursions, integrating exercise and stress data.
  • Multiple hormone delivery: Researchers are experimenting with glucagon or pramlintide in addition to insulin (a bi-hormonal artificial pancreas).
  • Cloud-based optimization: Some groups are exploring remote algorithm tuning based on population data, while preserving privacy.
  • Improved user interfaces: Streamlined mobile apps that reduce the number of devices needed.

If you wish to contribute, start by joining the OpenAPS Gitter chat, where developers and users discuss issues daily. Read the Getting Started guide, and consider setting up a development environment with the openaps-dev toolkit. Even if you are not a coder, you can help by improving documentation, translating materials, or simply spreading the word to clinicians and other patients.

Calls to Action for Different Audiences

  • For software developers: Pick an open issue labeled “good first issue” on the oref0 repository, or contribute to the RileyLink firmware repository to improve radio communication.
  • For hardware enthusiasts: Look into building a “hardware hacker” version of the RileyLink using an ESP32 and a CC1101 module. Share your designs.
  • For data scientists: Analyze the thousands of data sets available through OpenAPS users (with consent) to develop better prediction models for glucose.
  • For people with diabetes: Learn about the system, discuss risks with your endocrinologist, and consider whether DIY looping is right for you.

The OpenAPS ecosystem exemplifies how a motivated community, armed with open source tools and a shared mission, can create technology that changes lives. By contributing your skills—whether in code, design, testing, or education—you become part of a movement that challenges the status quo and puts control back into the hands of patients. The future of diabetes care is open, and it starts with you.

Learn more: OpenAPS Official Site | OpenAPS GitHub Organization | Nightscout Project | #WeAreNotWaiting Community