The MetroHealth System

David holds a BSE in biomedical and electrical engineering from Duke University, and MD, MS, and PhD degrees in biomedical engineering from Case Western Reserve University. He is well-known for his extensive work in informatics research, particularly focusing on electronic health records, personal health records, health information exchange, telehealth, big data, and clinical informatics education.

What are the commonalities between AI and patient safety in 2024?

There is tremendous potential in AI, specifically in generative AI. However, it also comes with cautionary tales. The notion isn't that AI or generative AI will replace humans, but rather that individuals who integrate AI into their work will outpace those who do not. Essentially, AI acts as a force multiplier, enhancing efficiency but also requiring supervision to be effective.

For example, in our generative AI pilot project, we're exploring ways to enhance personal health record responses. Traditionally, when a patient sends a query through their health record portal, the provider must read the message and compose a response. With our new approach, the electronic health record system generates a draft response based on the patient’s information, which is then reviewed and finalized by the provider. This aims to improve efficiency by reducing the time and effort required for crafting responses.

From a patient safety perspective, this approach leverages the extensive data in the electronic health record to generate responses. However, it’s important to remember that AI, as of 2024, is not flawless. Therefore, these AI-generated responses still require oversight and review by a licensed independent provider or physician to ensure accuracy and reliability.

Do you observe a particular pattern that companies should follow to form effective partnerships to adopt AI?

I believe that partnership is crucial. I’ve worked extensively with technology, and the value lies in the collaboration between the technology and clinical partners. While I'm embedded within the healthcare system as the clinical partner, it's essential to work closely with the technology partner. The technology partner brings cutting-edge innovations, but it's up to us on the clinical side to integrate these advancements in the safest possible manner. We aim to avoid regressing with technology in terms of safety, while also ensuring it remains productive and valuable for the healthcare system. The key to achieving this is to establish a strong partnership between the technology and clinical partners.

Could you share any interesting projects or initiatives you've been involved with at The MetroHealth System?

We're exploring how generative AI can assist with patient summaries in our electronic health record (EHR) system. We've been using our Epic EHR for nearly 25 years, resulting in some patients having decades of medical history stored electronically. For a new provider or specialist reviewing a patient for the first time, navigating through 25 years of information can be overwhelming. From the patient’s perspective, the expectation is that any relevant information is accessible to the physician in the EHR and the physicians are familiar with it. Generative AI could streamline this process by efficiently sifting through extensive data to create concise summaries. These summaries must be well-annotated so that physicians can trace back to the original data if they need to verify details or resolve any questions. This approach helps prevent misinterpretation and ensures the completeness of the information presented.

  ​The key to unlock technology’s full potential lies in dedicating time to understand how it can be used in healthcare to harness its true value for patient safety and operational efficiency.   

We're exploring the use of generative AI to improve the way we handle hospitalizations. When patients are hospitalized, numerous care providers are involved, and a lot of information is generated. After discharge, both patients and their families need a summary of the events that took place during the hospital stay. Generative AI could streamline this process by analyzing the extensive data from an inpatient visit and creating a draft summary for the healthcare team to review and validate. This approach could make the process more efficient and effective. From a patient safety perspective, AI could address the challenge of synthesizing extensive and complex medical histories, which might be difficult for individual providers to fully integrate, improving both safety and operational efficiency.

We've also extensively worked with predictive AI and predictive tools in our healthcare system to implement the sepsis predictive model and a no-show predictive model. Our sepsis predictive model, when integrated with the appropriate workflow changes, has successfully increased the number of days patients are alive and out of the hospital. Our no-show prediction model, combined with targeted interventions, slightly decreased the overall no-show rates and racial and ethnic disparities in no-show rates.

How do you envision the evolution of AI within the patient safety space in the future?

I’m optimistic about the potential of artificial intelligence in all its forms—generative, predictive, and rule-based. I believe that these systems will always need careful supervision, both from a technological and healthcare perspective. Understanding how they work is crucial to reap their benefits while avoiding issues like patient safety concerns and other potential drawbacks.

These advancements act as nudges that will enhance efficiency and patient safety, making incremental improvements rather than bringing about a dramatic transformation. While some people view generative AI as a revolutionary force in healthcare that will transform healthcare dramatically, I see it as more evolutionary—offering valuable contributions to patient safety and overall value, without fundamentally altering the landscape.

Any specific piece of advice that you'd like to share with your fellow peers or other industry leaders?

While the technology itself is straightforward, the real challenge lies in effectively integrating it into healthcare to create a meaningful impact. People invest considerable time and effort in developing the technology but neglect how it will be utilized. The key to unlocking the full potential of technology lies in dedicating time to understand its practical applications in healthcare to harness the value it offers.