How Cloud Services Are Enhancing Patient Care

Providers accelerated cloud adoption during the pandemic because it enabled them to quickly scale up their infrastructure to support patient care needs, including the increased demand for telehealth services, says Greg Caressi, global client leader and senior vice president of Frost & Sullivan’s healthcare and life sciences team. “Cloud got a huge boost from COVID-19,” he says.

But even before the pandemic, providers were already adopting cloud solutions. More than 50 percent of health IT workloads are currently deployed in the cloud, according to a recent Frost & Sullivan report.

Managing and maintaining legacy systems can cost up to 64 percent of IT budgets, so many organizations, particularly in healthcare, can’t afford to do everything on-premises anymore, says Mutaz Shegewi, IDC’s research director for provider IT transformation strategies.

“It’s mostly to modernize IT infrastructure and to facilitate digital transformation,” he says of cloud adoption.

Cloud solutions can also help reduce costs, break down data silos and improve data access across an organization or multiple stakeholders. For example, hospitals involved in a merger or acquisition can use the cloud to standardize their technologies and help integrate different aspects of their operations, Shegewi says.

LEARN MORE: Find out how else Providence adapted during the pandemic.

Real-Time Data Improves Care

Providence, a nonprofit healthcare system, partnered with Microsoft in mid-2019 to modernize its IT infrastructure with Azure cloud services.

When widespread pandemic restrictions hit in March 2020, Providence was in the midst of building a data lake and data warehouse in Azure by replicating data from its electronic health records, supply chain and other applications from its data centers.

Early in the pandemic, Providence tried to perform data analytics with its in-house data warehouse, but its servers couldn’t consistently handle the compute-intensive task simultaneously with daily data processing needs, so the organization quickly pivoted to Azure to power the effort, MacLaren says.

“We were hitting capacity constraints, so we felt this was a good first use case for the cloud,” he says.

Providence uses several Azure products and services, including Azure Data Lake Storage Gen2 for storage, Azure Databricks for analytics and machine learning, and Power BI to produce real-time reports on dashboards. The goals during the pandemic were to prevent hospital overcrowding and to forecast and anticipate staffing, PPE and ventilator needs across its hospitals, MacLaren says.

“The dashboards allowed our frontline caregivers and regional leaders to make critical decisions,” he says. “For example, if a hospital is full, could we divert incoming patients to other nearby facilities? Do we need to move ventilators to another part of the state that was hit harder? And where do we need additional doctors to treat patients?”

Dashboard use has waxed and waned over the past year, depending on infection rates, MacLaren says.

But real-time data streaming is just one use of the new cloud data platform. MacLaren is planning other data projects, including improving patient flow within hospitals, implementing cloud-based image archiving and deploying artificial intelligence to analyze medical images to detect diseases early.

“We are at the tip of the iceberg in unlocking the capabilities that the cloud enables,” he says.

The Benefits of AI-Powered Diagnostics

Researchers at UC San Diego Health in California developed a machine learning algorithm that analyzes chest X-rays and detects pneumonia associated with COVID-19 infections.

When pandemic cases soared in the spring of 2020, the university deployed the machine learning model on Amazon Web Services (AWS) under an approved experimental protocol to assist its experts with diagnosis. The algorithm overlays X-rays with color-coded maps that show the probability of pneumonia.

“When a patient comes in, the emergency room doctor may order an X-ray, and the AI results pop up next to the X-ray film within a few minutes,” says Dr. Albert Hsiao, a radiologist and associate professor at UC San Diego who developed the algorithm with his research team.

Physicians at UCSD Health’s emergency department need to make fast decisions on patient care, so they often have to review X-rays before radiologists can weigh in. The AI analysis helps support these decisions, Hsiao says.

“It’s ultimately the physicians’ ­interpretation of the X-ray and color map that matters. They can ignore it if they think the AI is making a mistake. But occasionally, the AI picks up something that would have been ignored, and it allows the physicians to look more closely,” he says. The algorithm has analyzed about 100,000 chest X-rays since last year, which is nearly all the chest X-rays performed at UCSD Health.

The university got the algorithm up and running within a week because UCSD Health previously built a secure environment within AWS, says Dr. Mike Hogarth, UCSD Health’s clinical research information officer.

To make it work, Hsiao and his team placed the algorithm in a virtual machine instance and connected UCSD Health’s clinical imaging system with the AWS protected environment the university had built. The benefits of the cloud are scalability and speed of deployment, Hogarth adds. It would have taken two to three months to purchase and install the data center infrastructure to run the algorithm in-house, he says.

“With the cloud, we had it running at the point of care within five days,” Hogarth says.

AWS initially powered the algorithm by giving UCSD Health free credits. Since then, Hsiao and his team have improved the algorithm. In December, they deployed the second version of the algorithm through the more sophisticated infrastructure of Hsiao’s private company. That also runs on AWS.

So far, the new algorithm has analyzed about 50,000 X-rays as it continues to provide UCSD Health’s doctors and radiologists with analysis, Hsiao says.

DISCOVER: How can artificial intelligence improve patient outcomes?

The Hybrid Cloud Makes Data Management Easier

Luminis Health in Maryland is ­taking a hybrid cloud approach. It operates a ­private cloud using Nutanix ­hyperconverged infrastructure, but two years ago, the organization began adopting public cloud services where it made strategic sense, says Associate CIO Ron Nolte.

The system has adopted Microsoft 365 for email and office productivity. When it was time to upgrade its in-house Oracle enterprise resource planning software, it migrated to Oracle Cloud Infrastructure instead. Luminis Health is also deploying a cloud-based image archiving solution from a provider that will store the images in Google Cloud Platform.

The organization is not cloud-first, but its IT leaders say they consider cloud options when it’s time to upgrade or purchase new applications.

“It’s picking the right tool for the right job,” says Juan Rodriguez, IT director of Luminis. “Where it makes sense, we will move a solution from on-premises to the cloud.”

READ MORE: SASE offers security in cloud migration for healthcare organizations.

For example, Microsoft 365 simplifies IT management, while Oracle Cloud Infrastructure saves money because Luminis Health doesn’t have to buy new hardware. By not having to manage the Oracle environment, IT staff can focus on larger projects that the organization still needs to keep on-premises, he says, adding, “Ultimately, it’s doing more with less.”

Recently, many cloud c­ompanies have made their services more feature-rich and secure, so Luminis Health is more open to the cloud, Nolte says.

“Industry is developing more compelling and reliable options, be it Software, Infrastructure or Platform as a Service,” Nolte says. “Vendors are becoming more polished and providing more support, so there’s less risk and greater value.”