GE Healthcare has taken initial steps to integrate 'omics data into its Centricity electronic medical record system through an exploratory research project that is developing a genomics data analysis infrastructure.
Mark Dente, GE healthcare's chief medical officer for healthcare information technology, discussed the project last week during a panel discussion at the American Medical Informatics Association's Translational Bioinformatics conference in San Francisco.
The panel discussed several projects that are looking to integrate genomics data into EMRs. In addition to Dente, panel participants included representatives from the Electronic Medical Records and Genomics Network, the Pharmacogenomics Research Network, and the HL7 clinical genomics workgroup.
BioInform spoke to Dente after the conference to get additional details about GE Healthcare's genomics infrastructure development plans. The following is an edited version of that conversation.
During your presentation at AMIA, you mentioned that GE Healthcare is developing a genomics analysis infrastructure. Could you provide some more details about what you hope to develop and where those efforts currently stand?
What we presented at AMIA was a mix of our technologies that we have today like our EMR and our ability to have large datasets to do research against. [T]he genomics effort is where we are headed, [but] it is not a product [now and] it may never become a product.
What we are talking about here is the driving of personalized medicine and translational medicine. I am a biomedical, clinical informaticist ... and my claim to fame is to think about knowledge management and clinical decision support and how we can shorten the ... bench-to-bedside timeline, [which] is about 17 years for something to go from research to full adoption in clinical practice. Now you compound that with a part of medicine that most clinicians in clinical practice are [unfamiliar with]. They learn a little bit of genomics in undergraduate school [and] in medical school but how do you educate folks as to ... where the research is going? Finally, how do we deal with new knowledge repositories in medicine?
A lot of our industry is run on old technology. We've made a large investment on the technology side looking at services-oriented-architecture. We can put legacy systems ... and new technology into this new infrastructure and because is platform we can aggregate data across the institution and even across the community and do analytics on this data in our data warehouse.... this SOA architecture is a modern way of doing that. [We have a] joint venture [with] Microsoft [called Caradigm that is] focused around that and advanced clinical decision support.
The final leg is [the] genomics platform itself … One thing around genomics is that there needs to be a higher expectation on the technology's ability to handle large datasets. An SOA infrastructure allows us to be more flexible on the technical side of dealing with genomic information. [Also,] you really want to think about a genomic repository external to the EMR. That is my personal approach and how I will strongly suggest that we as GE will approach this. You do not want to clog up your operational EMR database with genetic data because it's just too large. [Also, because] its genetic data, we need to have a higher expectation of security. We have rigorous HIPAA and other internal standards of how we manage and keep private patient information and that will get ratcheted up in the future.
As you start to put data into a genomics database, we need to marry up the genomic data with the phenotypical data coming off an EMR. The real exciting part [is] we can start looking at the genomic data coupled with the phenotypical data in and a genomic analytic engine concept. With a analytics engine and the creation of algorisms to look for signal how do we start to think about running very targeted studies and [looking] for signals that suggest that these four hypothetical genes, [for example,] could be predictive of a [disease] state?
Will the final incarnation of your platform include any of the currently available open source bioinformatics tools and capabilities for genomic data analysis?
It's too early for us to know.
Would the platform be open source?
We are looking at more of an open application programming interface model than an open source model and I say that because our SOA platform in Caradigm will have that model ... and there will be open APIs that other people can write to this infrastructure. The genomics platform would be one module on this SOA platform.
The advantage of that is in an open API model, whoever wanted to develop on our platform would be able to leverage all the services that are embedded in that platform, [such as] security services and patient identification services instead of having to rewrite all that code.
At some point I look to our platform being open standards and open APIs but these are complex systems and these will be living systems that we will need to maintain for years so we would want to be able to make sure that we can support that going forward.
You've mentioned Caradigm. Can you provide some details about the arrangement between GE and Microsoft?
In December, GE and Microsoft created a joint venture [in which] we both contributed assets. Microsoft had an SOA platform called Amalga; we contributed our Qualibria platform, which was a co-development with Mayo Clinic and Intermountain Healthcare, and all of our knowledge management and decision support, and we also contributed a health information exchange solution. The new company name is called Caradigm.
Now, an institution might have open source genomic information, [and] the hospital is probably running a GE or EPIC or Cerner system as their hospital information system. The "aha" moment for us was [that] it's about the data. It's about aggregating the information, bringing that together and then doing analytics against that. Whether its genomic analytics or research for diabetes care, it's about taking and aggregating data and that’s what Caradigm's strength is going to be.
I realize that this is still at a conceptual phase but could you give me a sense for how long it would take to bring a complete platform — one that combines both the EMR and genomic data analysis — to market? What sort of internal procedures does it have to go through at GE?
I can't speak to the genomics aspect because this is still a research concept. I can give you a generic overview of how we do review processes. Generically, we have a very rigorous protocol on how we look [at] product development. It's basically a yearly cycle and there's something called a growth playbook and how we look at the industry and where we want to be and, bluntly, what regulatory requirements we have to have.
Looking more broadly at the market, do you see any groups that are developing similar platforms to what you are envisioning? And if so, how will GE carve out a niche for itself?
We don’t see a lot of it because this is new. What you see are silos. You rarely see somebody that has the breadth, the domain length, and the staff to take on such a broad challenge and yet that’s what's really needed to make this usable. If not you are going to have two siloed systems — a siloed genomics system and a siloed EMR system — and then you have to figure out how to interface. I think we are unique in that we are thinking this way.
The other aspect is that SOA architecture. I think we are really unique in that we are driving a services-oriented architecture. A lot of the industry's hospital information systems — not so much the genomic platforms because they are new so they are built in a modern way — [are written in the Massachusetts General Hospital Utility Multi-Programming System, which] was developed in the '70s. It's hard for those systems to consume these new platforms. So we have made a huge investment with Caradigm to build that SOA platform and I think that is where our carve-out's going to have to be.