Predictive analytics could yield next-generation decision support, says Philips Healthcare CMIO/CTO

frassica

First-generation clinical decision support has been plagued by poor uptake among physicians, largely due to its overwhelming nature and perceived lack of applicability to clinical practice. But predictive analytics, built into these platforms, could produce the next significant wave in innovation in pediatric care, according to Joseph Frassica, MD, chief medical informatics officer and chief technology officer of Philips Healthcare.

Frassica spoke about predictive analytics last week at a panel on innovation acceleration at the Boston Children’s Hospital National Pediatric Innovation Summit + Awards 2013. Vector caught up with him afterward.

Vector: How will predictive analytics and phenotyping intersect to improve decision support?

Frassica: Right now, we have decision support tools based on our current knowledge. They are based on many patients and the signatures that those patients showed as part of the development of their disease process.

What’s missing is how to take the knowledge that has always resided in the physician’s head―information about the last 10 patients that he or she saw, combined with data on 1 million patients―and couple those with specifics about how the current patient reacts. When you put those two together, it becomes a really powerful decision tool.

Vector: Can you describe some of the barriers to integrating clinical decision support at the point of care?

Frassica: The use of decision support for physicians is a challenge. Initially, when industry introduced decision support, systems included too many guidelines. In addition, many guidelines weren’t relevant, so physicians learned to ignore decision support because it didn’t apply to them or their patients.

Now, we are moving beyond that by using high-resolution data to create decision support that is relevant to the patient. This information needs to be delivered at the right time [of the care process]. If the physician is not provided with relevant data that are accompanied by ‘what to do next’, the data will be ignored. The physician needs the data and the supporting sources to inform or change suggestions for patient care.

This kind of decision support is next-generation. It’s where we think the world is going. Integrating high-resolution information and delivering it to physicians when they need it will make physicians open their iPhones for clinical decision support.

Vector: How many years will it take for this type of decision support to be integrated into pediatric medicine?

Frassica: It’s challenging in pediatric medicine because of the data desert. We are closer in adult medicine, but in general, medicine is very close to being able to utilize predictive analytics to create large data sets.