Rethinking IT infrastructure in healthcare

In yesterday’s blog post, I wrote about the interest and excitement surrounding artificial intelligence (AI) and machine learning at RSNA 2019. While the enthusiasm for AI in medical imaging (and healthcare in general) is palpable, at least one presenter at the Annual Meeting attempted to temper expectations regarding the technology’s immediate impact on our industry.

In an educational session titled, Preparing Your Radiology Practice and IT Department for Big Data and AI, Paul Chang, MD, professor of radiology and vice chair of radiology informatics at the University of Chicago School of Medicine, outlined how weaknesses inherent in existing healthcare IT architectures impede true execution of AI initiatives.

Our future is everyone else’s past.

Dr. Chang began the presentation by reinforcing the healthcare industry’s position as a laggard when it comes to IT implementation.

“We are 15 years behind other industries when it comes to IT,” he said. “Our future is everyone else’s past. Other industries have been leveraging AI and machine learning for years, but healthcare has yet to scratch the surface of what the technology can do.”

Insufficient IT infrastructures limit healthcare’s potential

The reason, according to Chang, lies in the way most IT infrastructures are architected in healthcare provider organizations. Specifically, he notes that healthcare has invested in expensive vertical systems (e.g. EMR, PACS, etc.) that are difficult to integrate and scale.

While the IT systems healthcare providers have put in place are necessary, Chang argues they are not sufficient to meet the needs of the industry going forward. One of the main reasons, he states, is because the architecture of these systems relies on humans to launch applications and integrate workflows.

Before healthcare can truly reap the benefits of AI, Chang believes a completely different type of IT stack needs to be implemented – one that doesn’t abandon existing IT investments, but augments them. Furthermore, this IT stack needs to improve the suboptimal human/machine workflow collaboration that exists today, enabling machines to handle many of the mundane tasks currently required by humans.

Criteria for a new healthcare IT stack

According to Chang, this new IT stack should meet the following requirements:

  1. It should address present and near-future healthcare requirements and challenges
  2. It should transition from merely storing data and information to optimally leveraging knowledge
  3. It should evolve from standard business practices to data/evidence-driven business intelligence
  4. It should have the scalable capability to consume and analyze large and complex data (Big Data) in an agile, frequently real-time, manner
  5. It should transition from expensive vertical scalability (i.e. where servers, CPUs and storage are upgraded) to cheaper horizontal scalability (i.e. where more inexpensive commodity servers are added to the infrastructure)

Interoperability and orchestration needed before AI

Chang continued by providing an analogy for AI in healthcare where he likened AI to a fancy new sports car.

“Before you can drive this new sports car you need two things – gas and roads,” he said. “In this analogy, gas is data interoperability and roads are workflow orchestration. Currently, in healthcare, we have neither. We need to drill for gas and build roads first.”

Chang then covered various IT infrastructure strategies that healthcare organizations can employ to improve interoperability. These strategies include the incorporation of an independent data repository or warehouse; leveraging edge appliances or state aggregators; and building your own Service-Oriented Architecture (SOA), application programming interface (API), and microservices infrastructure.

Each of these options has their own set of advantages and disadvantages. For example, while a data warehouse may help consolidate and normalize data, there is some latency involved, so it’s not great for real-time applications. On the other hand, an SOA architecture is great at ensuring interoperability and enabling real-time applications, but it requires a dramatic IT cultural change.

Chang has implemented an SOA IT infrastructure at the University of Chicago Medical Center and has witnessed the agility and scalability advantages it provides firsthand. However, he has stopped trying to insist that other providers adopt this model because of significant impact it has on IT departments.

Regardless of what approach you take, Chang says the ultimate goal should be the same for all healthcare providers.

“Your Big Data and AI hedge strategy should be to prepare your existing IT infrastructure to be able to feed and consume near-future, advanced-decision support agents, including those that are cloud-based,” he concludes.

Rethinking your infrastructure? To see how Hyland Healthcare can help you modify and enhance your existing IT stack, visit booth #4300 at RSNA19.

Ken Congdon has expertise in the healthcare technology industry and has been a contributor to the Hyland blog.
Ken Congdon

Ken Congdon

Ken Congdon has expertise in the healthcare technology industry and has been a contributor to the Hyland blog.

... read more about: Ken Congdon