Alan Herbert, a geneticist at Boston University, said that the natural tendency of companies and universities is to lock up data contained in electronic medical records or banked tissue samples. The databases become a unique resource that can be curated to generate revenue, so there is a huge disincentive to release data, even though such access would improve the quality of everyones research.
Science is usually done in "small groups that hold data in small silos," said Carol Kovac, general manager of IBM Healthcare. She said that multiple experiences in IT showed the advantages of open access for the general community and for IBM in particular. She cited Linux as a culture that "rewards people for contributing, not holding back."
But the problem is even more profound said Yochai Benkler, a professor at Yale Law School, because different entities want to use the data in different ways. Because research clinicians seek trends that might help them to determine what treatments would be best suited for patients with particular genotypes or lifestyles, they aggregate data from many individuals. For their part, physicians want to have on hand all the relevant data about one particular individual. Patients want convenience and privacy, and payers want cost efficiency.
Data systems are fragmented even within one institution, Herbert said. Radiologists pick information systems suited to their needs, and clinical and billing information systems are often separate.
Thus, there are strong drivers against the standardization necessary to share data, Benkler said. He said the situation was unlikely to change without government intervention or the rise of one dominant company, like Microsoft. "We have an underperforming sector with many competing standards, where all the cost-squeezing [and other benefits] happens at the institution," he said.
Kovac was more sanguine, saying that the standard for exchanging medical images, DICOM, had become widely adopted through collaboration, and people are starting to see it as a model.