How Can Open Source Projects Support Themselves in Health Care?

Andy OramHigh prices and poor usability hasn't driven the health care industry away from megalithic, proprietary applications. What may win the industry over to open source (in addition to the hope of fixing those two problems) is its promises of easy customization, infinite flexibility, extensibility, and seamless data exchange. As we will see, open platforms also permit organizations to collaborate on shared goals, which appeals to many participants.

But if open source projects can't charge hundreds of thousands of dollars for installation as their commercial competitors do, how will they pay their developers and hold together as projects? This article compares three major organizations in the open source health care space: the tranSMART Foundation, Open Health Tools (OHT), and Open mHealth. Each has taken a different path to the universal goal of stability.

The tranSMART Foundation focuses on pharmaceutical and biotech companies, as well as academic and nonprofit biomedical researchers; Open Health Tools on the health care industry and EHRs; Open mHealth on popular fitness devices and other sources of mobile health data. I'll come back later to these distinctions, which may well be critical to the choices they made. But all these organizations share the goals and values that run throughout the free software and open source community. They strive to build community, seek to build platforms that all participants can build on constructively, and emphasize an open process that puts the needs of diverse users above the interests of any single contributor.

I have focused on the development process and organizational philosophy of the tranSMART Foundation in another article. Its main take-aways are that the founders of the tranSMART Foundation have treated it as a start-up company, just like any other business, ensuring its funding and operations before taking on projects and bringing in other managers with a strong background in the business of healthcare research and software engineering.

Open Health Tools did not take this route--and perhaps were never given the opportunity to do so. They see themselves as a clearinghouse and promoter for worthy open source projects in areas such as data creation and exchange, leading eventually to a shared platform for electronic health records (EHRs) and e-health environments. Although specific projects were funded and OHT received targeted grant monies, these never reached a sustainable level. They have just found stability, really for the first time, by joining the HL7 standards group, as one of their working groups.

Open mHealth started with grant funding, expanded into providing software, and now has a hybrid model.

To learn about these organizations' strategies and what they have recently accomplished, I talked again with CEO Keith Elliston of the tranSMART Foundation, and interviewed both Ken Rubin, the Interim Executive Director of OHT, and David Haddad, Executive Director and Co-Founder of Open mHealth. tranSMART and Open mHealth seem to be firmly on the path toward their goals, and OHT is enjoying a new lease on life that is too early to evaluate but whose results a we will see over the next year or two.

Keith Elliston, Ph.D., is the president and CEO of the tranSMART Foundation.tranSMART strengthens its foundations

According to Elliston, the tranSMART Foundation spent more or less its first year focusing on standing up the business and getting its funding in place. The Foundation spent its second year focusing on building a critical mass in community and on executing on a number of key deliverables. The success of this second step emerged as more than 165 people attended its third annual meeting, held recently at the Netherlands Cancer Institute (NKI) in Amsterdam.

To expand the platform deeper into the industry--among both current members and new users--the next phase for the Foundation is to improve the tranSMART platform from its current research grade to a true commercial grade. This is a manifestation of its philosophy that open source projects must be as appealing and easy to install and use as commercial offerings. (It is worth noting here that many people attribute the spectacular success of MySQL to its ease of installation and low barriers to getting started.) Several of the projects taken under the tranSMART Foundation's wing emerge from academic settings and stressed functionality over ease of installation and ease of use.

Their task therefore becomes adding user interfaces and installation procedures in order to make the platform more user-friendly. For instance, the original tranSMART required 54 manual steps to install. Now they have a scripted installation that automatically handles all the dependencies. It takes about three hours, but requires minimal manual effort. The tranSMART Foundation has also developed a new roadmap that calls for regular six-month releases similar to the Ubuntu distribution of Linux. This means that the release process takes six months, and a six-month development process on the next version goes along in parallel. This results in a well defined set of releases that come out at a predictable rate. This provides a stable deployment base that can dramatically expand the user community.

The tranSMART Foundation also realizes that for open source projects to become adopted by the mainstream, they need to be robust and reliable. This is a function not just of software quality but also of configuration, testing, support, and distribution.

The second major aspect of this phase is to expand the funding base of the Foundation. To this end, they're hiring their first development manager. This role is focused on expanding the funding of the Foundation’s effort beyond their current membership program, into philanthropic, grant, and sponsorship funding. To enable these two efforts, they have expanded their staffing in two key areas: software engineering and fund-raising.

Even with the new personnel, the tranSMART Foundation remains very lean. No one works full time. Elliston says they look for people with extensive entrepreneurial experience and a track record of success in business, who now want to take on a philanthropic project. The staff are expected to be skilled, experienced, and perhaps most importantly, passionate about the mission. Each person on the payroll works somewhere from 30% to 75% of their time at the tranSMART Foundation, and the management teams boasts over 100 years of combined C-level experience in entrepreneurial businesses.

Ken Rubin, Interim Executive Director of Open Health ToolsOpen Health Tools offers infrastructure to thin applause

At the outset, Open Health Tools tried to establish themselves through a three-phase approach:

1. The first stage, "storming," attracted like-minded organizations and individuals working on e-health projects and convene them under one community.
2. The second stage, "forming," standardized the community's governance, culture, and goals as part of a transformation from a chaotic group into more of a like-minded community. It took about two years to establish this culture, and this period was the span when OHT showed the most promise and the most community interest.
3. Finally, the third stage, "norming," was intended to standardize around a core architecture and e-health platform upon which products, systems, and capabilities could be built. Open Health Tools never successfully achieved that third stage, largely because potential funders didn’t share OHT’s understanding of the value of shared infrastructure.

The task of designing an open source platform might be compared to the SMART and FHIR projects. Those open source projects focus on data exchange among electronic health records, and now work together. Open mHealth, described in another article, focuses on medical and fitness devices. These organizations came along much later than OHT, and perhaps had more measured aspirations.

Within the classic health IT sector, vendors are busy building stand-alone products with focused business models, while health care providers are focused on squeezing value out of their current software and meeting regulatory requirements such as Meaningful Use and the ICD-10 transition. On a more fundamental strategic level, health IT is seen throughout the sector as a cost, not an investment. Thus, participation in OHT had been declining over the past couple years. "There was a 'sweet spot' opportunity to help organizations achieve initial Meaningful Use certification, if the platform had been in place at that time, but it wasn’t," says Rubin, who has been involved in OHT since the mid-2000s. "Missing that window really hurt our ability to be sustainable."

Rubin said that a couple organizations came through early in OHT's history to offer a firm funding base. But there were too many strings attached. Essentially, these organizations wanted OHT to become a development group to meet their idiosyncratic business goals and optimize its tools for them. The organization would also have had to abandon the egalitarian goals of openness, transparency, and meritocracy that are key to the open source movement. The offers were rejected.

In contrast, when HL7 and OHT began discussions, the nature of the relationship was fundamentally different, and a much better fit for the long term. HL7 is inherently more sympathetic to what OHT does. And in fact, HL7 has been a member of OHT for a long time. By taking it over, Rubin says, HL7 gives OHT enhanced visibility and legitimacy. HL7 sets standards for software development but has no software development team of its own. So OHT can bring in valuable resources. Furthermore, HL7’s interest in open, standards-based systems and data sharing align perfectly with OHT’s vision.

Another key piece of the environment that pulled OHT and HL7 together was the creation of FHIR. Although it started as a private skunkworks project by several independent developers, HL7 quickly recognized it as the fast path to a modern standard worthy of 21st-century software development, and adopted FHIR as a standard. This in turn creates hopes in everyone's minds for a third-party ecosystem of software tools that work on all electronic health records. OHT should have a role to play here.

All very nice in theory. Will OHT take off? We'll have to wait to see. In January HL7 will hold a working group meeting, one part of which will introduce OHT to its larger community. OHT is also in regular touch with FHIR leaders, but has not yet held discussions about how precisely to work with them, and how FHIR relates to a potential OHT platform.

David Haddad, Executive Director and Co-Founder of Open mHealthOpen mHealth creates a hybrid model

Founded in September 2011 with an explicitly open source mission--to create a common standard for mobile health data so it can be used in clinical environments--Open mHealth was hatched successfully through funding from two sophisticated foundations with roots in health care, the California Healthcare Foundation and the Robert Wood Johnson Foundation. This seed money allowed Open mHealth to create its first API along with an app to show it off.

This success led to a bit more funding: some $500,000 that allowed Open mHealth to show how the open standard and API can be used to integrate more than just one app together. At a 2013 Convergence Summit, Open mHealth received lots of enthusiastic praise. But nothing came of it. They realized they needed to appeal to the immediate needs of their potential community. As Haddad said, the explicit goal went from creating an open standard to "making life easier for engineers." 

Thus, preparatory to creating the Shimmer data aggregation tool (highlighted in another
article, Open mHealth created four or five mappings between various fitness devices. They made choices based both on which organizations had relationships with them and which devices were currently popular.

Recently, Open mHealth has invested more in community-building. This included their blog (where the "making life easier for engineers" outlook is obvious), creating MeetUps that currently have over 700 members worldwide, holding its first summit that maxed out with 120 attendees, and hosting clinical measures groups where the organization can present their schemas and get feedback. The organization plans to do more online and offline community engagement on different continents.

At the same time, Open mHealth has built an organization to conduct training and consulting around their work. Because this could become a distraction and take attention away from their core open source mission, they are exploring a new company that would handle these types of contracts.

Cultural differences set the course

Why did tranSMART and Open mHealth find a firm grounding, with an active community and adequate funding, while OHT floundered for years? Part of it was planning and timing; the leaders of the tranSMART Foundation and Open mHealth managed to intuit what their potential communities wanted and how to stay on track. But we also have to examine the different cultures they occupied: biomedicine and consumer fitness versus health care providers.

Biomedical researchers and companies are natural risk takers. They envision massive, disruptive changes from things such as genetic research. One out a hundred companies may succeed, but the success stories will be big ones, as suggested by Barry Werth's 1995 book, The Billion-Dollar Molecule. Consumer fitness is also a fast-growing market with great promise, where the participants know they are at the very outset of an adventure.

In contrast, health care providers have developed a culture of caution derived from decades of negative lessons. Conservative procedures, followed compulsively, prevent errors in patient care. Innovations should also be eyed cautiously, as shown by Otis Brawley in How We Do Harm when he retells the history of eagerly adopted treatments that turned out disastrous. When this caution extends to information technology, however, a lot of useful innovations get passed by.

I can also speculate, hearing of the attempts by various organizations to co-opt OHT in its early years, that the health care industry didn't appreciate the multiplicative effects of open source development and just didn't "get it" in general when it came to collaboration as an driver for innovation.

Still, all three organizations face some of the same challenges. For instance, unlike the web sites who create and use many JavaScript tools and cloud technologies, the consumers of these organizations' software are less likely to create them or actively contribute to their code. As Elliston says, tranSMART ties together pharma companies who possess a lot of money but little development expertise with researchers who possess a lot of development expertise but little money. OHT was in a similar position during its rise.

In a situation like this, the core open source value of forming communities is crucial, and both organizations understand that. Elliston says "The Art of Community is our Bible." (It so happens that I edited that book.) Software, funding, and community are the three legs supporting the open source stool. Over time we will see the particular ways tranSMART, Open mHealth, and Open Health Tools strengthen each leg.