Aidoc announced an agreement with Northwell Health to enable an enterprise-wide clinical AI strategy across 17 of its New York state hospitals. In a strategy to further enhance its system-wide healthcare, Northwell is leveraging Aidoc’s AI operating system (aiOS™) to implement AI across service lines with the goal of driving patient follow-up after discharge to connect the dots of care.
We recently spoke with Dr. Matthew Barish, Vice-Chair, Informatics, Radiology Service Line at Northwell Health to discuss AI’s place in Northwell, barometers for measuring its success and Aidoc’s value in furthering Northwell’s commitment to patient care.
Aidoc: I want to talk a little about Northwell’s vision. The words “innovation” and “strategy” are becoming commonplace in health systems, and especially when it comes to a hospital’s approach to technology. You’re in the field of imaging informatics, so what is your vision for Northwell in terms of your approach to technology and how did AI become a factor?
Dr. Barish: We look at technological advancements as toolkits for us to utilize and improve patient care and provider experience. AI fits well into that framework where it has patient care implications without adding an additional burden to our providers. In fact, we’ve found AI reduces the overall burden for providers. Unlike most EMR processes, where you’re improving care of patients but burdening providers, AI, especially in imaging, can really impact both the experiences of patients and providers by decreasing the amount of time it takes to review a study and prioritizing studies. It’s a win-win for both parties. Lastly, it has a positive impact on revenue and you can justify ROI for the investment in AI that gives back either an actual hard dollar value to the radiology department or infrastructure of the enterprise or, by decreasing time, you can get efficiency gains.
Aidoc: That’s an interesting point you bring up about ROI, because it’s still a conversation being fleshed out with AI. What convinced you that AI has the potential to bring ROI and downstream benefits to Northwell?
Dr. Barish: I think that the hardest part is understanding that there is value at the end. The only way you can know is to start your AI strategy by clearly defining the value you’re seeking and how to measure it. Typically, you will implement a tool like AI to solve a problem, and the problems you’re looking to solve surround patient care and provider issues where you want to improve performance or sensitivity. But it’s unusual to think of an AI tool as a revenue generator; that’s not the first thing people think of when they try to solve a problem. But it’s something you have to think about when considering an AI strategy. So during the implementation process itself, we made sure we set out parameters to measure ROI:
We look at whether we could discharge patients on the basis of an AI result, especially in an outpatient setting, thus having our technologist and staff at the outpatient centers focusing on the next patient because they no longer need to focus on a patient waiting for results. These are things we looked at and measured from day one so we could determine whether we achieved our ROI goals.
Aidoc: I want to ask you about the story of the enterprise AI strategy and how the aiOS supports that from a technical perspective. Can you share how this approach resonated at Northwell?
Dr. Barish: When we made a decision that AI would be a key technological differentiator, we wanted to be a market leader in the deployment of useful, clinically impactful AI that has a positive ROI. We wanted to find a partner that had the concept of a platform understood, where we integrate once and can feed multiple products through that infrastructure. We looked through vendor platforms that gave the promise of that, but they don’t necessarily deliver a lot of their own AI. When we understood what it meant to have point solutions, you essentially end up deprecating the effect of the AI solutions that you deliver. There would always be something you gave up. What we quickly realized is that there were vendors like Aidoc that had a wide breadth of algorithms that met the commonly experienced problems in radiology. It may not solve all of them, but they solve enough to know it would deliver a better, all-encompassing, and sustainable solution for AI. We’re already delivering third-party applications through Aidoc at Northwell and plan to do more so having that aiOS infrastructure makes it easy to do so.
*This conversation has been edited for length and clarity.