electronic health records
apixio

Apixio’s New Iris Platform Uses Your Doctor’s Notes To Derive Insights

Next Story

LG Is Developing Its Own Mobile Payment Service

Data science applications for healthcare are finally trying to catch up to the rest of the world, with one new effort coming from six-year-old Apixio in San Mateo, CA. This morning, the company is launching a cognitive computing platform called Iris that derives insights from clinical data and other information in the health system.

More specifically, Iris uses a powerful data integration and extraction suite and pipelines restructured data into a machine learning model that should, in theory, enable its healthcare provider customers the ability to better understand patient health.

It’s easy to draw an immediate parallel between Apixio and IBM’s Watson Health system, though there are critical differences. Watson derives insights from medical literature to correlate symptoms with outcomes, whereas the Iris platform uses doctors’ records and notes.

 

Comprehensive Data Integration [19179469]

While data-driven techniques in medicine are hardly new, Apixio’s target market has historically been one with little widely accessible information. Though doctors produce volumes of notes about patients, handwritten and otherwise, most of this data is siloed in healthcare records and unavailable even in newer electronic forms. What is accessible is difficult for machines to read and rarely indexed.

“Eighty percent of the medical records are being unused by technologies that are being leveraged in healthcare today,” says Darren Schulte, Apixio’s CEO.

Bob Rogers, one of Apixio’s cofounders and now Chief Data Scientist at Intel, adds that part of the problem is the typical doctor’s workflow. “Turns out doctors don’t use codes for their own work,” he said. “They dictate or write to themselves and to other doctors.”

Apixio’s innovation is in being able to operate around this workflow. Its operation centers on its ability to access undigitized data, then parse, and derive powerful insights from it through natural-language processing techniques tied to machine learning algorithms.

It’s a big bet. As Schulte explains, it’s up to customers to decide whether they want to share data or metadata to help the machine learning system derive better insights. In short, as innovative as the potential insights of Iris could be, the system could be easily hamstrung by large healthcare providers who are more restrictive of data.

The good news: Over the years, Apixio has built up a powerful network of large healthcare providers, from Scripps Health to HealthNet and WellCare. All have the kind of rich data sets that are critical to data science applications that can help Iris fulfill its potential, provided they continue to share them.