CHICAGO (GenomeWeb) – Partners HealthCare in Boston has long been at or near the forefront of health IT advancement. Its Massachusetts General Hospital set up the world's first telemedicine system 50 years ago and was an early adopter of electronic health records.
In the last 15 years, the health system has gotten a jump on what is now called precision medicine by building up its genomic informatics infrastructure.
"That infrastructure includes lots of different elements," noted Samuel "Sandy" Aronson, executive director of IT at Partners Personalized Medicine.
"There are systems to support laboratory workflows, making sure that our physical assessment of samples is robust and as high-quality as possible. There are systems that manage the interpretation process," Aronson said. "There is a system that helps us make sure that, as new information emerges on variants that were previously identified in a patient, that clinicians are alerted if that information could impact that patient's care."
Partners actually registered the interpretation infrastructure as a medical device and began distributing to other organizations in the US and Canada under the GeneInsight brand name. In April 2016, the health system sold GeneInsight to Sunquest Information Systems.
Since then, Partners has refocused its IT innovation work on an app-based infrastructure called Health Innovation Platform, or HIP, to address individual workflow needs as they arise. It goes beyond early genomic IT infrastructure work that concentrated on genomic interpretation.
"Now, what we are focused on doing is expanding what genetics and genomics is — at times a component that needs to be combined with other forms of data to help inform those process flows," Aronson said.
"The goal of [HIP] is to make it easier to build the kinds of IT interventions that are needed to alter process flows," Aronson said. The basic HIP infrastructure features security and audit-log protocols, as well as calculators that can be reused in new apps, he noted.
"A great deal of energy associated with building these apps involves data integration. The platform provides a mechanism where you can hook up underlying data sources," Aronson said. Integrations generally follow the Fast Healthcare Interoperability Resources (FHIR) standard, including the FHIR Genomics flavor of it.
"What we're finding is, as we work in different clinical areas, the reuse of those adapters across clinical areas is more significant than we had anticipated," Aronson said.
This app-based strategy generally involves coding applications that put existing data sources into new workflows within the security and integration framework of HIP. Programmers code the "user experience" into each app, then release the software into the clinical environment, according to Aronson.
For example, after sequencing exomes or genomes, a laboratory would scan each sequence to look for variants, reducing the size of the initial data set for easier interpretation at the clinical level.
"Some of those variants will be interpreted, and will be clinically relevant. Those will get deeply interpreted if they are part of the indication for the test. [There is] much deeper information about each variant that's found, but far fewer variants," Aronson said. "That data set is what, in our environment, actually gets exposed into the clinical infrastructure or labs that are hooked up to expose it."
Of course, each department has its own data requirements.
"The lab needs to be able to see the stack of the raw data that indicates the likelihood that a variant is actually present. The clinicians may not need to see that," Aronson said. "Different data is used in different ways, and that affects how the infrastructure gets built."
Partners recently launched an app to address the rejection of platelets in bone marrow transplants, which is generally caused by a poor human leukocyte antigen match between marrow donor and recipient.
"It turns out that the data that we need to deal with that proactively either exists in the hospital system or can be readily generated, so you pull together information on the patient's HLA profile, the donor's HLA profile, the patient's platelet count, [and] the platelet units that are in inventory at a given point in time," Aronson explained. Adding a calculator to perform analytics from the main HIP infrastructure, and Partners had an app to reduce the chance of platelet rejection.
"You do wind up building a platform that is intended for general use, but for each individual component of the platform, you make sure that it's built in the context of providing some real-world clinical value," he said. "By doing that, you make sure that it's appropriately validated and robust."
Aronson said that the HIP strategy addresses the "last-mile problem" of bringing data, analytics, and genomics-driven clinical decision support to the bedside.
"The HIP platform is our strategy for addressing that last-mile issue and actually making it possible for us to deploy the IT capabilities that are needed to actually get new innovations into use as quickly, efficiently, and safely as possible," he said.
To date, HIP has only been deployed in house, but Partners eventually intends to release the technology to the open-source community, Aronson said.