Tapping big data

Doctors might be notorious for writing illegibly, but they still take a lot of notes. Like, a lot of notes. In fact, every year doctors, nurses and other health care providers in the U.S. take about 2 billion of them, says local entrepreneur Kyle Silvestro—entering information about patients into computers, scribbling something by hand for later transcription, or dictating with speech-to-text software.

Silvestro calls those notes “unstructured pieces of information,” and he says somewhere around 80 percent of health care data takes such a form. He sees that deluge of data as a vast and largely untapped resource that could be used to help guide outcome-based care.

The problem is, it’s not like Silvestro, or anyone else, can sort through each note individually.

“We have a human problem,” he said during a recent interview. “We can’t throw enough labor at this, to read every single document coming through and take that information and enter it into a database so you can actually do research on it.”

That’s where Silvestro’s Chico-based company, SyTrue, comes into play. His firm has developed software to sift through unfathomable volumes of data and provide it to clients, such as hospitals and insurance and pharmaceutical companies, so they can draw their own insights on physician performance and patient outcomes. In other words, he’s pushing “data-driven health care.”

“Clients use us to get information and they use it for things like predicting whether somebody is going to end up in the intensive care unit or go home,” he said.

Indeed, while SyTrue can provide a sweeping view of the health care industry from above, Silvestro says his software has applications at the point of care, as well. As an example, he recounted taking his daughter to be seen by a doctor the morning after she fell hard on her elbow one night. She had an X-ray taken of her arm, and Silvestro saw the radiologist examine the image and document his observations.

But when they left the hospital around noon, they were told the report wasn’t ready yet, and they’d get a call later that day.

“Six hours went by, so I called them and asked what was going on, and they said, ‘Oh, your girl has a broken arm,’” Silvestro said. “I asked when they knew this, and I got excuses. … I kept pressing, and I found out that document was ready to go at 11:52 a.m., before we even left the hospital.

“Our technology has the ability to read that report and identify that she had a broken arm and that it wasn’t communicated.”

That’s something SyTrue helps to address with its client vRad, the nation’s largest network of radiologists, which includes professionals in 2,200 hospitals nationwide.

“In radiology, we’re identifying something in a document and making sure that is actually communicated to the patient,” Silvestro said. “So they don’t need to wait too long for that interaction to occur.”

And since vRad is such a broad network, it’s able to use SyTrue’s technology to determine national trends and averages, such as how often X-ray scans discover something noteworthy like a broken bone or tumor.

“What percentage of time are you actually finding something?” Silvestro posed. “Are you over-utilizing services patients don’t need?”

Since launching SyTrue in 2012 out of an office on Morrow Lane, Silvestro, a Chico State graduate, has slowly grown the company from a one-man operation to one that now employs 12 tech workers living all over the world, from Canada to Armenia. And he believes the ever-evolving niche SyTrue occupies has serious growth potential because, so far in the ongoing digital-data revolution, the health care industry has been slow to get on board. But Silvestro says that’s changing.

“We’re really just starting to hit that curve,” he said.

SyTrue is on the forefront of big data in health care for two reasons, Silvestro says. First, it’s just plain difficult for software to account for the variations in the way humans communicate—especially misspellings, abbreviations and differences in syntax and documentation. Due to the challenges of developing such software, there aren’t many other companies with rival technologies, he said. SyTrue’s software uses natural language processing, a kind of artificial intelligence that, in basic terms, makes the computer understand data as a human would.

Second, there wasn’t much incentive for the health care industry to draw insights from data prior to the recent, industry-wide shift toward outcome-based care now emphasized under the Affordable Care Act.

“Doctors have been paid by the visit,” he said. “If you’re a doctor, I go to your office and how much time you spend with me determines how big of a check you get. Now you have a shift in health care that’s saying, ‘I’m going to pay for outcomes.’”

Moving forward, Silvestro hopes SyTrue can help “solve some of the toughest problems that health care faces,” such as streamlining the process for doctors taking in new patients.

“If you want your next physician to go through your medical record and make a decision, they have to read through every page of a thick booklet. It just doesn’t make sense. If you can have a technology that can go through and extract all the pertinent information, you can get a whole history of someone in a glance, rather than spending five hours of reading that file.”