Cloud computing can provide new possibilities for powerful, flexible, and cost-effective collaboration and innovation in medical research and health care, but there are some pretty big dark clouds hanging in the way.

That seems to be the consensus among researchers and technology leaders who met with representatives from Amazon (NSDQ: AMZN).com and a small handful of other vendors this week at a forum in Boston to explore the role cloud computing can play in the biomedical and health care fields.

At an invitation-only event sponsored by Harvard Medical School and Amazon Web Services, a few dozen experts convened in Boston for a day to ponder the possibilities of cloud computing in their work. Participants included health care IT leaders, academics, biomedical researchers, medical and scientific consulting firm representatives, and officials from vendors like Amazon, Oracle, and Hewlett-Packard.

Because of its elasticity, scalability, pay-as-you-go model, and other characteristics, cloud computing can potentially provide huge cost savings, flexible high-throughput, and ease of use for resource-strapped biomedical researchers who need to collect and crunch terabytes of complex information, like human genomic data, in the pursuit of medical discoveries.

In addition, Web-based servers, storage, databases, and other cloud computing infrastructure, software and services also offer an attractive platform for collaboration among medical researchers across the globe, as well as for public-health officials across the United States.

Despite the cloud's allure for heavy-duty medical research and data-intensive health-related applications, there are big hurdles standing in the way. For one, government regulations regarding privacy and security make putting medical data up in the cloud risky if there's any chance the data, for instance about genomic or health issues, could somehow be tracked back to specific patients.

Researchers and tech experts at Harvard Medical School, Partners HealthCare, and Children's Hospital Boston, for instance, are investigating or in early development of specific research and other applications that tap into the collaborative, flexible nature of the cloud.

Ken Mandl, a researcher and physician with roles at Harvard Medical School and the informatics group at Children's Hospital Boston, is involved with ongoing development of several public-health surveillance applications for Massachusetts and at the national level. The cloud could, for instance, provide a flexible platform for public-health departments to upload health data in a timely manner to assist state and national health officials in the early identification and tracking of disease outbreaks, environmental-related health problems, and other issues, said Mandl.

But many questions persist right now -- such as whether public health departments can legally allow patient data to reside on the cloud, how to standardize applications used for those purposes, whether vendors like Amazon and others would allow "their part of the cloud" to be used for public health purposes, and data security and privacy issues, Mandl said.

"Health data is always in a special category," said Mandl. "Laws around it are different, and there are many special interest groups to protect it."

At Harvard Medical School, the laboratory of personalized medicine is already using some cloud-based services, including Amazon Simple Storage Service (S3) and Amazon Elastic Compute Cloud (EC2) for translational science and simulation research work.

"It's like a virtual lab," said Peter Tonellato, senior research scientist at Harvard Medical School's Center for Biomedical Informatics. The platform "fits the vision of ubiquitous access to the lab on the Web regardless of location." Using the platform, researchers can do "cool analysis" of clinical and genetic data using "clinical avatars," or simulated representations of patients.

"Clouds are here to stay," said Tonellato, who predicts that many research organizations will "transition" to private/public cloud infrastructures for elasticity and cost-efficiency in their data analysis work.

The cloud can also provide a resource for collaboration and knowledge-sharing in data-intensive research and analysis, especially in the health and biomedical arena, said Jill Mesirov, chief informatics officer at Broad Institute. Mesirov described her organization as "a genomics center on steroids," because of the multiple petabytes of genetics data the institute's ''infrastructure collects, analyzes, and archives."

The cloud provides a platform that can help "incorporate prior knowledge" among the work of biomedical researchers across the country, as well as an infrastructure that supports "dynamic scaling" for the variable demands that complex genomic data analysis requires, Mesirov said. "It's very expensive to run and maintain all that equipment. The idea of sharing this with others is compelling," she added.

For its part, Amazon in recent weeks unveiled the AWS Hosted Public Data Sets, or "Public Data Computing Initiative," which provides on the cloud a "hosted-for-free, centralized public repository" for data -- such as United States census and human genome research data -- useful to researchers, said Adam Selipsky, VP of Amazon Web Services.