Health Catalyst’s analytics effort aims to improve treatment decisions
Health Catalyst is hoping to revamp the approach for using data to improve care and make better cost-based decisions by incorporating more types of data and enabling a closed-loop analytics process that pushes findings back into electronic health records systems.
The Salt Lake City-based vendor on Tuesday announced what it’s calling a Data Operating System, which combines features of data warehousing, clinical data repositories and health information exchanges.
The concept will use open APIs to better support third party application development, says Dale Sanders, executive vice president of product development for the company.
“For the vast majority of organizations, health information exchanges, traditional data warehouses, personal health records and EHRs have all failed to make the necessary information for decision support available at the point of decision,” Sanders says.
It’s important for an analytics and data warehouse platform to be able to incorporate a wide range of data sources and use those in developing a treatment plan, particularly as providers move toward precision medicine and population health. Sanders says only a tiny fraction of the data required for these emerging care approaches reside in current EHRs.
The Health Catalyst approach uses a hybrid transactional analytical processing approach, believed to be the next iteration of data warehouses, the architecture of which will need to evolve toward a broader data management solution for analytics.
Health Catalyst’s Hadoop big data approach will provide a separate “layer” in IT operations that makes it easier to work with data. It draws from so-called Kappa Architecture, which allows for one stream for batch and real-time computations in the serving layer. The architecture can be implemented with a combination of open-source tools like Apache’s Kafka, HBase, Hadoop, Spark, Drill, Spark Streaming, Storm and Samza.
Sanders says the machine learning output from the new system can be fed back into the EHR through the use of the Fast Healthcare Interoperability Resources (FHIR) specification being developed by HL7. “Because it’s FHIR based, it can send findings back to the workflow, whether the clinician is accessing the EHR on a computer or mobile device or through some other kind of mobile application.”
The company reports it’s invested $200 million to develop its Data Operating System. It will enable healthcare organizations to better capitalize on its local content within its data stores; in addition, the data will be aggregated by Health Catalyst in a repository that will enable research by any of its clients.
Two of Health Catalyst’s clients already are using the data operating system, and Sanders says the company plans to roll out other customers on the system in the coming months.
The approach will enable healthcare organizations to use any combination of products that make sense for them, enabling them to take advantage of emerging analytics and data approaches from small developers who are attacking various healthcare challenges.
“I think it will enable application development like I’ve never seen before,” Sanders says. He noted that a variety of industry pressures have caused providers to eschew best-of-breed approaches and select vendors that offer a monolithic approach to clinical systems.
“A monolithic model has its benefits but also has drawbacks,” he adds. “This swings the pendulum to what we’re used to as consumers. On my smartphone, each application has a valuable specific purpose to me, and I would never want a single vendor to address all those needs. In healthcare, we really had no other choice until now, and we believe the new system will enable developers to write applications about 10 times faster than before. We believe it will open up what can be written, and it can extend the life of existing EHR systems.”