(Editor's Note: This is part two of two of this interview. To read part one, click here.)

Dr. Bruce Darrow, chief medical information officer and interim chief digital and information officer at New York's Mount Sinai Health System offered some thoughts this week about why artificial intelligence is having such a big moment in healthcare and how AI may one day be taking some (emphasis on "some") cases away from doctors.

In this Q&A, Darrow offers a closer look at how Mount Sinai is using AI – and how it plans to expand its use. He discusses how long the health system has been using AI for clinical care, what principles its clinical and IT leaders follow when considering clinical AI use cases and the AI deployments Mount Sinai has in place today. In the video accompanying this article, Darrow also describes what determines whether an AI initiative at Mount Sinai is likely to succeed.

Q. How long has Mount Sinai been using AI for clinical care, and where has the health system been using it?

A. You might think the use of AI is a more recent development. It depends on where you draw the line and where you make the definition.

We've been using algorithmic care and ways to use computer-based decision support for many, many years. The first real application of AI was back in 2013. It's been more than 10 years at Mount Sinai, where at that point, the first use case we reported or published on was using AI algorithms to find patients in the hospital who were likely to get very sick before they got to that point and their outcomes were worse.

And by using AI to find them earlier in their care, we were able to improve their likelihood of surviving their hospitalization by a significant amount. So that's been more than 10 years.

And a lot of the work we've done at Mount Sinai over the past 10 or 12 years has been in the area of what I'd call predictive AI, finding patients who are likely to get sick, finding patients who are likely to have a condition that would benefit from having that knowledge, bringing the right skillset, bringing the right expertise, bringing the right treatments to that patient's care earlier in the process.

In the past year or two, we've been looking at ways to use AI for just streamlining care, not necessarily related specifically to clinical care, but ways to make care easier for our patients, to streamline the operational parts, as well as to start to automate some of the things doctors and other members of the healthcare team do that take a lot of time that can be drafted and the pre-work done for them.

Q. What principles does Mount Sinai use when considering clinical AI use cases?

A. This is very important. As I said, we've been using AI for more than 10 years, and we found over the past two or three years, as it became clear AI would be a growing portion of our patient care portfolio, that we needed to be purposeful about how we would go about using it.

At Mount Sinai, the principles we latched onto were that the use of AI for clinical care should be safe, effective, equitable and ethical. Safe and effective, obviously, we have to have tools that make a difference in a patient's care. They have to work. They have to be in the service of some goal that advances care.

Ethical and equitable in terms of the way we make sure we are bringing those tools to all of our patients in a way that aligns with our mission as an organization.

Q. What AI use cases does Mount Sinai have in place today?

A. Most of the AI we use comes from basically three different pipelines. We're fortunate at Mount Sinai to have a very talented and engaged team of data scientists, implementation scientists, artists and other team members who can use a learning platform, a data pipeline, to make our own AI algorithms, test them and use them for the care of our patients.

They've published extensively and been recognized for this. David Rich, the president of Mount Sinai Hospital, and Robbie Freeman, who is our chief nursing information officer and vice president within Digital and Technology Partners for Innovation, have been very active with their teams.

Some of the examples are finding patients before they get sick enough to need ICU care, identifying with greater accuracy than existing tools whether a patient in the hospital is likely to be at risk for falls, identifying patients who are at risk for malnutrition or pressure ulcers so we can bring it to the attention of the right members of the care team.

These are great supplements to the care our nurses, our doctors, our social workers, our registered dietitians are already providing in the hospital setting for our patients.

We have a lot of homegrown knowledge and expertise, and we've been doing that basically since about 2016. In the last five years or so, we've seen a growing amount of imaging AI. These are all FDA-approved tools and software algorithms we can use for our patients.

Many of these do not, as I said in yesterday's discussion, replace the radiologist or the clinician, but they make that radiologist's work more accurate, more efficient, faster. One example is if you imagine there may be 20 patients who have had head CTS, that's a computed tomography of the head, to look for abnormalities that could include a stroke or bleeding within the head.

If a doctor is looking at a list of 20 of them, he or she may not know. They may go in order of when the images were acquired. But if you have AI running in the background and it says, out of these 20, look at these two first, because these are the two that are likely, according to the algorithm, to have something that looks abnormal. That's good for the clinicians.

They get their attention to the right studies first, and it's good for the patients because they get their care faster when we think it may make a difference in their care. There's a fair amount of imaging AI for both diagnostic accuracy and just making sure we have the right selection of where the attention should be given.

Then the third area where I see a lot of AI is in the tools provided by our existing software or other software providers in the community. Just about every piece of software we use at Mount Sinai, if it doesn't already have AI built into it, I can expect it to have AI built into it over the course of the next 3-5 years.

It's just the way that technology is going. Our electronic health record system has embedded AI we consider and validate and decide whether or not to bring into care. Just everything from email to presentation documents to video collaboration we use is going to have some element of AI in it.

BONUS CONTENT: Click here to watch a video of this interview that also includes Dr. Bruce Darrow discussing what determines whether an AI initiative at Mount Sinai is likely to succeed and what his peers at other hospitals and health systems can take away from this.

Editor's Note: This is the ninth in a series of features on top voices in health IT discussing the use of artificial intelligence in healthcare. To read the first feature, on Dr. John Halamka at the Mayo Clinic, click here. To read the second interview, with Dr. Aalpen Patel at Geisinger, click here. To read the third, with Helen Waters of Meditech, click here.

To read the fourth, with Sumit Rana of Epic, click here. To read the fifth, with Dr. Rebecca G. Mishuris of Mass General Brigham, click here. To read the sixth, with Dr. Melek Somai of the Froedtert & Medical College of Wisconsin Health Network, click here. To read the seventh, with Dr. Brian Hasselfeld of Johns Hopkins Medicine, click here. And to read the eighth, with Craig Kwiatkowski, senior vice president and CIO at Cedars-Sinai, click here.

The HIMSS AI in Healthcare Forum is scheduled to take place September 5-6 in Boston. Learn more and register.

Follow Bill's HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.