It's staggering how little we know about disease. Pharmaceutical companies sink billions into drugs that attack the wrong targets. Researchers in labs study fragments of the big picture gathered from patients--often just a few hundred--entered in clinical trials. Combining those fragments so that they represent tens of thousands of people would create much fuller models of illness and speed the production of effective drugs. But in a world of competitive grants and publish-or-perish academe, data hoarding is the norm.
Meanwhile, people in pain stay that way longer.
Stephen Friend wants to change the process and culture of biomedical research by liberating that hoarded data, coaxing patients to offer even more information about themselves, and providing virtual collaboration space and heavy-duty analysis tools for scientists to build more complete pictures of disease together. Sage Bionetworks, the Seattle-based foundation Friend co-founded in 2009, is building a vast technology platform called Synapse, on which researchers worldwide can deposit the raw data they collect on patients and take out the data deposited by others. They can then work on that data individually or in groups, performing complex pattern searching and other mathematical operations with the help of Synapse's software and scientific staff.
The results of their work then go back into Synapse, for others to build on. In essence, Sage is trying to replace the siloed structures of academe, biotech, and pharmaceuticals--what Friend calls the medical-industrial complex--with an open-source model. "Building these models of disease is going to take decades and multiple groups," says Friend. "Anyone who suggests they're going to get to the truth with a solo effort is just scraping the surface of the data they need."
Friend has been working at the nexus of Big Data and disease since the mid-'90s. That's when he co-founded Rosetta Inpharmatics, a company whose products analyzed torrents of genetic data. After selling Rosetta to Merck in 2001, Friend stayed on as head of oncology research and developed two intriguing partnerships. One was with a Florida cancer center that had amassed rich banks of data on patients. The other was with a company in Iceland, where the government maintains genealogical and medical records on every citizen. "I thought, 'Having this much data works really well,' " says Friend. "And having much, much more data would work better."
Friend has raised $30 million from industry, government, and foundations. Partners including Merck, Pfizer, Johns Hopkins University, and groups focused on Alzheimer's and Huntington's disease primed the pump by contributing clinical data. Another source of data is patients themselves. Many ill people are happy to release the results of their tests if they believe it will hasten new treatments. Some will cheerfully heap the plate with detailed medical histories, environmental factors, even their genome sequences.
The chief obstacle to such patient participation has always been the legal consent forms created to protect privacy. "People who run studies often don't want to have to explain the complicated reality of what might happen if your data is made available," says John Wilbanks, a director at Sage and fellow at the Ewing Marion Kauffman Foundation. "It's easier to get people to agree to very simple consent forms that say we will never share this information with anybody." Wilbanks is developing a new type of form--called Portable Legal Consent, or PLC--that gives patients the authority to donate their own data. For example, 500 patients who share a rare disease could join forces and compel attention from the medical community. "They might say, 'We are going to create a really rich data resource about ourselves and then put up a $50,000 challenge, and the first person to build a model of our disease gets the money,' " says Wilbanks. In addition to increasing the power of patients, the PLC will reduce the time and money wasted by researchers, who can reuse data and conclusions from other studies rather than constantly generate their own.
The pharmaceutical industry stands to save even more. Drug companies spend $4 billion to $11 billion to bring a new drug to market, much of that squandered on repeated failures. Research and development departments direct drugs at targets--enzymes or receptors--they believe will produce certain results. The work done on Synapse is intended to deliver detailed disease maps with You Are Here arrows pointing at the best targets. For the industry, "the cost of bringing products [to market] has been growing exponentially, while productivity, measured by new drugs on the market, is going the other way," says Thomas Krohn, director of clinical open innovation at Eli Lilly. "Sage is doing some very interesting things that could accelerate the timeline for getting drugs to market."
Sage doesn't raise significant competitive issues for pharmaceutical companies. Things are more complicated in what Friend calls the precompetitive space of basic research. Grants and published articles feed the careers of scientists and researchers, who naturally want to keep their findings under wraps. So Friend is experimenting with incentives to get people to use his information commons. "Until we find new rewards, no one is going to share just because the technology is there," says Friend.
So, for example, Sage enlisted five research organizations, including Mount Sinai Hospital, Columbia University, and Stanford, to spend a year collaborating on projects about aging, cancer, and diabetes. The participants--normally vigorous competitors--made significant progress and jointly produced several papers. Sage is also inviting patient advocacy organizations to deposit data into Synapse and then build disease models on the basis of the aggregated information.
Each spring, Sage hosts a congress attended by users and observers. Friend has capped the number of attendees at 200 and turns away thousands. "People are starting to say this is the way the world is going to be," says Friend. "And I want to be a part of it."