Discussion with Dr. Eric Poon from Duke Health Regarding Integration of AI and Introduction of Copilot Technology
A Chat With Duke Health's AI Pioneer: Dr. Eric Poon
Healthcare tech is abuzz with AI's positive impact on clinical workflows, improving patient outcomes, and boosting productivity. But finding the right solution that delivers results while fitting the organization's budget and gaining necessary buy-in can be a challenge.
That's where Dr. Eric Poon comes in, the chief health information officer at Duke Health, who has been at the helm of various AI implementations, including ambient listening and Microsoft Copilot. Here he shares his insights on finding the right AI tools, embracing the 'fail fast' mentality, and choosing ambient technology that's bringing real change in Duke's health system.
HEALTHTECH: What are some AI implementation challenges healthcare organizations face?
POON: Learning that not every technology or AI solution will work for each organization is crucial. A range of factors could cause this, such as technology or organization readiness, or whether end-users are adequately prepared.
At Duke, we believe in embracing a "fail fast" mentality. You need to pick your frogs wisely and decisively, as not every frog will befit a royal role. But when you find a promising solution, prepare the organization to embrace it fully and swiftly.
Ambient technology has been an excellent example for us. Over the past few years, we've explored various vendors, iterated implementations, and aligned ambient tech with our workflow. After conducting a head-to-head trial comparing two leading vendors, we've found amazing results, creating organizational buzz, and achieving widespread adoption with over 1,200 providers actively using it daily after the initial rollout.
EXPLORE: Here are 13 ways AI enhances healthcare operations, patient care, and treatments.
HEALTHTECH: What foundational technologies, infrastructure, or policies do healthcare organizations need to have in place to support AI initiatives?
POON: Having the right decision-makers in place is essential to identify solutions that meet organizational needs, test them out, and ensure the technology is delivering intended results. Measuring success is a priority, as letting go of unpromising solutions is vital for any organization.
Building a prepared workforce ready to embrace AI technologies also plays a significant role. We leveraged the mass rollout of Microsoft Copilot to encourage early adopters and provide support for those who wished to use the technology.
DISCOVER: Three ways Microsoft's Copilot in Windows can help productivity.
HEALTHTECH: Speaking of guardrails, what security controls need to be in place before jumping into AI use cases?
POON: Patient privacy is an utmost concern in healthcare. We've put multidisciplinary teams in place to determine appropriate use and weigh in on partners, ensuring the technology is secure enough for protected patient data.
We also developed guidelines for generative AI in clinical use cases to create a safe and responsible environment for AI adoption throughout our organization.
Dr. Eric Poon, Chief Health Information Officer, Duke Health
HEALTHTECH: How else do you recommend healthcare organizations go about preparing for AI implementation?
POON: Measuring the impact of technology is vital, but traditional IT governance often fails to do so effectively. For AI, stakes are high, and we're still in the early days, so it's essential to set success and exit criteria for technology solutions. This helps decide whether the technology is worthwhile and enables the organization to act accordingly.
By allocating the required personnel and time to measure technology outcomes, healthcare organizations can determine the best investments in times of limited resources.
READ MORE: Understanding AI Evaluation and Implementation in Healthcare
HEALTHTECH: How can healthcare organizations measure the ROI of their AI implementations throughout the lifecycle of those solutions?
POON: ROI measurement is crucial, but demonstrating ROI upfront during a pilot can be challenging. It's essential to invest time and effort in measuring ROI over time.
For instance, we knew ambient technology would improve satisfaction and decrease burnout initially, but to gauge increased productivity, retention, and other long-term benefits, we committed to measuring these outcomes for over a year.
HEALTHTECH: Is there anything you want to add?
POON: Identifying the right group of clinical or end-users to serve as "frog kissers" is essential for testing and validating AI solutions. A diverse spectrum of perspectives, personalities, and workflows will help represent a majority of users, ensuring swift and effective implementation when promising technologies are identified. Nurturing a group of innovators is a valuable investment.
- SCIENCE: With Dr. Eric Poon's expertise in AI, he has been instrumental in adopting technologies such as ambient listening and Microsoft Copilot, which are rooted in science and can revolutionize healthcare workflows and improve patient outcomes.
- TECHNOLOGY: Embracing technology like ambient technology, backed by rigorous testing and head-to-head trials, can bring about significant change in health systems, enhancing productivity, and facilitating seamless patient care, ultimately leading to improved health-and-wellness outcomes.
