CHICAGO (GenomeWeb) – This month, healthcare analytics technology vendor Cota announced a partnership with Memorial Sloan Kettering Cancer Center in New York to build clinical and genomic data sets in an effort to improve patient care and research.
Under the five-year deal, Cota will take anonymized clinical records from MSK and run it through its Cota Nodal Address classification system to create patient cohorts by categorizing patient-specific medical histories, information on their diseases, and planned therapies. This, representatives from the both cancer center and the company said, will help MSK make sense of vast stores of unstructured clinical data and prepare for the new world of value-based payments.
"We decided to utilize [Cota's] services based primarily on their fit for organizing unstructured data, which is a primary need of ours, to just access this wealth of information we have in the electronic health record, and ultimately couple it with genomic profiles and other data," said Paul Sabbatini, deputy physician-in-chief for clinical research at MSK.
"A lot of the data we measure here is very objective and very easy and structured," Sabbatini said. "But what's missing is all of that unstructured story [about each patient] in the electronic health record."
MSK is among the first Cota customers to rely so heavily on genetic testing.
Sabbatini noted that MSK has sequenced the genomes of more than 20,000 patients and adds 150 new ones a week. "There has been a lot of manual augmentation of the genomic data along the way, but what we really want to do is get a systematic way to marry all of that clinical data that's in the EHR with the genomic data," he said.
"We want to couple that with the genomic data we have here," Sabbatini explained. "If we could just get the clinical data and pair it with those things, we could probably make some very important observations to ultimately improve outcomes for people. This is the missing piece."
MSK already uses some natural-language processing technology as well as some search functions to help clinicians and researchers get at "semi-structured" data, according to Sabbatini. "But, there's just a lot of prose in the medical record. What most people are having to turn to is some sort of machine-assisted abstraction, which also requires a clear human element to get the data out," he said.
"It just became apparent to us, as it is to most institutions now, that if you really want this data, in the very near future, NLP alone will not extract the data elements in the EHR the way it was previously constructed," Sabbatini said.
"Hopefully, we're getting smarter about how we are putting the data in, and this won't always be true, but we have literally thousands of patient records where we know the genomic characteristics. We really want to get the clinical characteristics and match that."
New York-based Cota — formerly known as Cancer Outcomes Tracking and Analysis — has some staff dedicated to chart abstraction, but its primary selling point to MSK was Nodal Address, which creates numerical representations of patient characteristics. "We actually can pull out [and] read the doctor's note, and understand what it means, and then translate that into a structured data set," said Stuart Goldberg, chief scientific officer of Cota and a practicing oncologist himself — specializing in leukemia— at John Theurer Cancer Center in Hackensack, New Jersey.
Cota is able to tap into EHRs to extract both structured and unstructured data, then cross-reference it with patient-specific genomic information. "That's the thing that Sloan Kettering really found intriguing," according to Goldberg.
With the genomic database MSK has now, researchers can look for genome-based patterns in cancers and treatments. "The genome people can be very focused on what they're doing, but without adding that clinical outcomes piece, they don't know what they're looking at. That's where Cota comes in because we can pull both sides and marry them," Goldberg said.
"This is not just retrospective looking. We want to be able to tell [clinicians] that, [for] the people they are actively treating today, what they can expect the outcomes to be for those patients, [and] how they compare to other patients across the country so that they can help the patient that's in front of them," Goldberg continued.
"We're hoping that's going to really ... provide insights that no one else in the world could see without this combined clinical-genomic database."
Indeed, it is a two-way partnership with Memorial Sloan Kettering feeding data to the vendor. "We also want to give them the ability to benchmark what they're doing against national big data sets," Goldberg said. "Adding a partner such as Sloan Kettering for us elevates that benchmark dramatically. It really helps in developing a national database to have a premier institution such as Sloan Kettering as part of that."
Cota will be deidentifying MSK's data sets and offering access to others, including pharmaceutical companies. There have been some "preliminary discussions" with such partners, Goldberg said.
Another aspect of the relationship is the link to value-based healthcare.
According to Goldberg, this helps differentiate Cota from other analytics and cancer database companies. "We understand that we're moving into a value-based world, and you can't just look at outcomes. You have to look at the cost of those outcomes," he said.
Goldberg cited partnerships Cota has with payors, including Horizon Blue Cross Blue Shield of New Jersey, an investor in Cota. "We are actively trying to figure out how do we integrate the cost side into this so that we can present to the physicians the value piece, because that will be increasingly important as we move into modern medicine," Goldberg said.