HIMSS24: Health Care AI Guardrails Get a Boost while Cloud-Based Apps Soar

HIMSS24: Health Care AI Guardrails Get a Boost while Cloud-Based Apps Soar.  A guardrail with white and red stripes is in the foreground.  An artificial intelligence AI brain hovers over the open left hand of a man in a business suit while storm clouds gather in the distance.

Amid the nonstop news coming from the recent HIMSS24 tech conference, there was a decided shift in how many in the field perceive and use artificial intelligence (AI). And make no mistake, AI again overwhelmingly dominated content and conversations at the event.

Issues like how tech companies and provider organizations can work together to accelerate optimal uses of AI, the need for guardrails to ensure the ethical and responsible use of AI and how cloud-based systems can be used to improve diagnostic efficiency and accuracy were punctuated with real -world illustration cases.

And although there remains a long way to go as health care strives to harness and implement AI’s potential, here’s what resonates with us.

Takeaways on AI’s Present and Future

1 | AI guardrails are coming.

While there is increasing commitment to ensuring that AI deployment in health care produces credible results, the framework to help this happen is still under construction. The good news is that more groups are focused on this issue. The Coalition for Health AI, which began in 2022 but was formally launched last month, seeks to harmonize standards and reporting for health AI and create a common testing framework that all stakeholders can use to report results. John Halamka, MD, president of Mayo Clinic Platform, the health system’s big data initiative, recently was named to chair the coalition, which has 1,300 members.

Meanwhile, Microsoft recently joined 16 health systems to launch a stakeholder group to implement AI guardrails. The Trustworthy & Responsible AI Network will share best practices and create a process for members to register their uses of AI in clinical settings to enable outcomes measurement with the technology. Among the initial participants are Stanford Medicine, Johns Hopkins Medicine, Mass General Brigham, MedStar Health and Advocate Health Care.

Takeaways

It is significant that health systems are helping to lead efforts to develop AI guardrails and best practices for the development and deployment of AI applications. Keeping patients, including their families and communities, as the focus of attention ensures that best practices are established from the outcome with equity and fairness, while ensuring that opportunities for traditionally underserved communities are embedded.

In health care, lives are at stake, and there is a responsibility to make sure that the benefits of AI and any innovation outweigh the risks by advancing a set of technical measures and metrics that can be used across a variety of use-cases with community consensus.

2 | More health systems have their heads in the cloud.

It was clear from partnership announcements at HIMSS24 that strategic AI applications through the cloud are driving greater use of the technology. Over the past two years, companies have been testing hundreds of thousands of generative AI proofs of concept, Matt Renner, president of North America and Global Startups at Google Cloud, told Fierce Healthcare.

Takeaways

While most applications don’t make it into production, some that do have great potential to address health care’s greatest challenges.

For example, Hackensack Meridian Health is using AI to help physicians detect advanced kidney disease sooner, which extends the time before dialysis treatment is needed, and even the need for transplants, system CEO Robert Garrett said during a keynote speech.

HCA Healthcare is working with Google Cloud on generative AI tools to streamline nurse handoffs. The health system developed a virtual assistant to gather patient handoff data throughout the day so it could be provided to nurses on the next shift. With 20,000 daily handoffs among nurses from one shift to the next across the system, it took anywhere from 60 to 90 minutes to document this activity on paper, Renner said. The virtual assistant automatically generates handoff reports and is designed to promote continuity, consistency, patient safety and clinical quality — while saving nurses significant time and maintaining human oversight.

Meanwhile, the enterprise software developer symplr is working with Amazon Web Systems to develop AI assistants to help automate complicated workflows and repetitive tasks for workers and to power chatbots to help patients find and schedule the right care.