KnowMED walked away the big winner of the Semantic Start-Up Competition. The Semantic Web Blog caught up with CTO Matthew Vagnoni, MS, and CEO Jerry D. Scott to further discuss the company’s winning entry, the Clinical Discovery Platform, for helping the health care sector semantically integrate data and ask natural language questions of that data, to support clinical research and complex decision-making.
The problem that the health care industry at large faces of not being able to easily and efficiently integrate and share data across organizations’ borders is equally a challenge right within the institutions themselves. “Large modern health care organizations are somewhat insular,” says Vagnoni.
At Medical City Children’s Hospital in Dallas, as an example, there are three separate electronic health record systems just for its neonatal division. The diverse formats and vocabularies made it difficult to try to ask questions of this data for research or efficiency purposes. But within two months of deploying KnowMED’s Clinical Discovery Platform, Vagnoni says, most of the data was integrated into a single view, “so clinicians could interact with it almost like using Google. …We combined the data from all the different sources so that clinicians could go in and ask questions [that reflect] how they think, not how [the information] is in the data schema.”
That’s helping forge the path for clinicians to more readily reach the ah-ha moment that comes when previously unnoticed data relationships are exposed. It promises to make it easier for a physician to make the connection between, for example, a certain medication that one patient is taking for one disease that actually is helping her survive another disease longer than expected, without all the manual labor and cost overhead that otherwise goes into making that happen. “Taking data arising from different places with slightly different meanings, and understanding their common meaning and common reference point – that’s what the semantic web lets us do,” says Scott. "That's what differentiates us from customary approaches to 'map' data from a source to a destination."
The platform’s pedigree includes patents on natural language processing (NLP) for parsing unstructured text in electronic health records and other sources, identifying key terms, determining context, and assigning meaning; converting text into RDF and making it queryable; vocabulary management; and dynamic form generation to help researchers speedily link to data outside clinical systems to further help with their information probes. It leverages W3C standards like SKOS, RDF and Direct Mapping of Relational Data to RDF, to drive data interoperability across diverse systems and give clinicians graph-based data representations for more dynamic interactions. “You don’t have to redo what’s been done before with the semantic web,” says Scott, “in contrast to relational databases, which adds to costs and makes them brittle.”
There is lots of data already collected in electronic health records, says David Booth, senior software architect at the company, “but it is currently very hard and time-consuming to extract the knowledge and insights from it that are needed.” KnowMED’s tools allow regular clinicians or researchers, not just database experts, to rapidly query and analyze their data in new ways, he says, adding that the company also “is a big supporter of the idea of using RDF as a universal healthcare exchange language.”
Booth, who also was one of the leaders of the RDF as a Universal Healthcare Exchange Language workshop held at SemTechBiz, says that’s something with a benefit for everyone – patients, clinicians, researchers – “because it would make healthcare information more interoperable. That would improve the efficiency and effectiveness of healthcare while reducing costs.” KnowMED's technology, he says, makes it easier to connect and make use of healthcare information, something that’s needed whether or not RDF is used as a universal healthcare exchange language, “but the widespread adoption of RDF for exchanging healthcare information would remove a lot of the friction.”
State of the Startup
In addition to Medical City Children’s Hospital, the one-and-a-half year old KnowMED has as a customer MD Anderson Cancer Center. “Our initial focus is large institutions who see value in being the best,” Vagnoni says. They want both to be able to show consumers the lead they’re taking in research, and from there, find new cures. “All hospitals see their data as important intellectual property, but all of them also struggle to make use of that intellectual property. The problem in health care isn't getting data into these systems, but getting data out.”
There are some 150 or more large Academic Research Centers in the country, and beyond that there’s also opportunity in community and rural hospitals. Vagnoni thinks there’s a role to play servicing them in the future via a secure HIPAA-compliant cloud solution. Cloud support for the long-tail of healthcare delivery organizations that desperately want to participate in research and need to leverage their data for government mandates is also high on its priority list, but slated to be rolled out later.
The more immediate focus will be on continuing to improve the platform’s NLP capabilities, and mature its terminology services around healthcare terminologies like SNOMED and ICD-9.
KnowMED also has partnership agreements in place, such as one with Afoundria, to support patient health population searches in its Integrated Health Services Environment. That’s a workflow platform for coordinated and connected healthcare, and the partnership means that clinicians can query clinical data for ad-hoc research and quality assessment, and other reporting needs, to smooth transitions in and out of post-acute care and reduce readmissions. “How do you manage populations and prevent them from readmission, how do you segment different groups and see about interventions, like calling or following up with an institution if patients are getting infections from a certain place,” Vagnoni explains. It’s becoming critically important for clinicians to have a strong grasp on evaluating patient states as the government begins to tie health care payments more to managing patient conditions and diseases and less to reimbursements for performing specific procedures.
When it comes to winning more health care organization clients, Vagnoni explains that it can’t focus KnowMED demos on the technology per se. “The decision makers are doctors with decades practicing medicine,” he says. The way to get their attention is to show them what the technology can do in the way of “more grants, saving more lives, and driving better and more efficient care.”