HLTH23: Microsoft unveils end-to-end data, analytics platform as hospitals look to get deeper into AI

In May, Microsoft rolled out Microsoft Fabric, an end-to-end, unified analytics platform, to lay the foundation for artificial intelligence applications.

As the pace of tech innovation in healthcare accelerates, the tech giant is rolling out new tech solutions to help healthcare organizations wrangle their massive troves of data and turn it into insights.

On Tuesday, Microsoft unveiled industry-specific healthcare solutions in Fabric that will enable organizations to combine data from previously siloed sources such as electronic health records (EHRs), picture archiving and communications systems (PACS), labs and claims systems and medical devices.

Available in preview, the healthcare data solutions in Fabric help to eliminate the costly, time-consuming process of stitching together a set of disconnected, multimodal health data sources like text, images and video.

The solutions bring structured, unstructured, imaging and medical device data into the Fabric data lake with open data standards using FHIR, DICOM nad medtech services, to provide one common architecture, Microsoft said.

Fabric also provides a multimodal data foundation to help organizations build standardized, scalable solutions to produce clinical and operational insights, the tech giant said. Organizations can then leverage this data to build and run AI models.

The company also launched new AI capabilities for clinicians and researchers to more quickly surface important information.

Hospitals produce 50 petabytes of data per year, or equivalent to approximately 10 billion MP3 music files, according to the World Economic Forum. The majority of that data, about 97%, goes unused, Alysa Taylor, CVP, Azure and Industry at Microsoft wrote in a blog post.

Healthcare organizations are challenged by data fragmentation, data velocity, with information moving much more rapidly, privacy and security concerns, a range of data formats and a lack of time and resources to harness all this data, said Umesh Rustogi, general manager at Microsoft Healthcare Industry Cloud 

"What these organizations need is a data strategy and a set of technology to help them unify all this data together," Rustogi said during a Microsoft preview event for reporters held Sunday during the HLTH conference.

'One of the biggest problems we have in healthcare, and Northwestern Medicine is not immune to it, is data is everywhere," Doug King, senior vice president and CIO at Chicago-based Northwestern Medicine told reporters Sunday. Massive hospital M&A activity in the past decade has only exacerbated this problem, he noted.

"Having that data in all those different areas makes it very difficult to really try to get real-time data insights to patient care or to hospital operators to understand what is happening holistically with that patient. It's challenging to work with the data and to leverage the power of what it can do within healthcare," he said.

As health systems move further into the age of AI, the need to harness that data becomes more critical, he noted. "AI needs curated data sets that are consolidated; it's incredibly difficult to train high-quality algorithms when you have data in different spaces," he said.

"Real-time situational awareness is becoming more critical when we start looking at healthcare. We have an aging population, we have high burnout, being able to have that data to impact the patient care and help a provider is incredibly powerful," King said. "If you have a new patient trying to get in to see a doctor and an appointment is not available for 10 days, this can trigger an alert for a practice manager to move things around to get that patient in sooner; that's powerful. That impacts patient care."

But data and analytics solutions in healthcare also need to have strong cybersecurity controls.

"Cybersecurity is the number one thing that keeps me up at night and probably most CIOs across the country," King said. "The Fabric solution, leveraging real-time FHIR data, and being a secure environment, it's a solution that we're looking forward to in the future."

King said Microsoft Fabric could be a "game-changer" for Northwestern and the healthcare industry overall. "If you look at healthcare historically, we have been behind the bell curve in adopting technology and behind in being progressive in that space. That's been with intention as we take care of patients."

By consolidating and harnessing data and using powerful analytics tools, healthcare organizations can not only meet the bar set by other industries but "set new bars," King said.

"Every CIO in the country doesn't want to meet the bar that somebody else has set. They want to leapfrog that. That's what we're hopeful for to truly leapfrog and transform the way we are engaging patients, the way providers are engaged and the data we can use to make better decisions at the time of the point of care," King said. "Having all that data there, real-time insights, to be able to change a care plan for a patient based on a predictive algorithm that is going to be very powerful going forward."

More advanced data and analytics tools also set up healthcare organizations for compliance with new regulatory requirements around data exchange and interoperability, he noted.

Arthur Health is working with Microsoft to use Fabric to power predictive care stage models and SingHealth, Singapore's largest network of public healthcare institutions, is using Fabric's healthcare data solutions for its underlying data infrastructure.

On the AI front, Microsoft rolled out Azure AI Health Insights as a service that provides prebuilt models to perform analysis. The company launched three new models in preview. One model, patient timeline, uses generative AI to extract key events from unstructured data like medications and organizes that information chronologically to give clinicians a more accurate view of a patient's medical history.

Another model, clinical report simplification, uses generative AI to take medical jargon and convert it into simple language to be shared with others. Radiology insights provides quality checks through feedback on errors and inconsistencies.

New capabilities in Azure AI Health Bot also integrate generative AI into healthcare chatbots and virtual assistants, the company announced.

"We talk about patient experience, we talk about burnout, we talk about how we can leverage these things; data is king and it is foundational to all things we want to do in all of those areas, patient care, provider burnout, what's happening within the healthcare institution to increase throughput or decrease length of stay so we can see more patients, data is in the center of all of it," King said.