Cedars-Sinai’s newest department, Computational Biomedicine, is up and running, giving the Cedars-Sinai Newsroom an opportunity to speak with the department’s founding chair, Jason H. Moore, about the history of this relatively new field, how to mine big data and the future of artificial intelligence in healthcare.
Newsroom: What is computational biomedicine?
Moore: Computational biomedicine is a discipline that brings together mathematics, statistics, and importantly, computer science and computer technology, to address all the important questions in biomedical research and healthcare that we’re interested in.
That’s really what computational biomedicine is about: bringing all those disciplines, the technology and the human side of this all together, so that we can transform healthcare using research results.
Newsroom: Tell us a little bit about the history of computational biomedicine.
Moore: I’ll point to 1977, which was a big year for my field of computational biomedicine. There were three big things that happened that year. The first was the home computer came out—the Apple II and a couple other home computers came out—and were available in stores for people to purchase. The internet was demonstrated for the first time in 1977. And on the biology side, DNA sequencing was invented, and the first DNA sequencing technology that was available for people to use came out just after that.
So this convergence of home computers, the internet, and the ability to generate a lot more biological information about DNA sequences was really the nucleus of when my field took off. And over time, data increased, computing power increased, the use of the internet increased. And that really gave rise to the field of computational biomedicine, allowing us to use computers, to use the internet, to manage and analyze and make sense of large volumes of both experimental data coming out of biology research labs, but also clinical data coming from hospitals.
Newsroom: How can computational biomedicine help improve patient care? Are there examples of how this field has already improved healthcare?
Moore: Patients benefit tremendously from computational biomedicine in so many ways. One of the big things I would point to is the electronic health record that we use to store and manage patient data in hospitals around the world. The field of computational biomedicine developed the first electronic health records back in the 1970s and 1980s, when computers came on the scene, and we started thinking about ways to manage the clinical data being collected to get it off the paper and into the computer. So that’s one of the biggest areas where the field has impacted patient care, because that allows us to much more efficiently capture all the information that we need about patients, to provide that very rapidly and easily to clinicians—doctors making decisions.
One of the best examples that we have right now is the use of artificial intelligence—computer algorithms—to interpret retinal images in the eye that are captured for diagnosing diseases like diabetic retinopathy. Computers and computer algorithms— artificial intelligence—can do a better job of diagnosing diabetic retinopathy from retinal scans than ophthalmologists can. And that technology has been FDA approved in clinical practice and is being used today to diagnose patients and diagnose them earlier and more accurately, so we can catch it and treat it before the disease gets too severe and does damage to eyesight.
The third example I would point to is a hot, relatively new area called drug repurposing. We have tons of FDA-approved drugs for specific diseases. But some of those drugs can be repurposed to treat other diseases. And that’s really exciting because once a drug is FDA approved, we know it’s safe. We know patients are not going to be at high risk for adverse reactions. And so computational biomedicine can play a very important role in analyzing the data that can help you figure out whether drugs could be repurposed for another disease.
Newsroom: It seems like these types of projects require a lot of data. How much data is required for these types of calculations?
Moore: Well, a home computer or laptop computer holds about a terabyte of data. And many of us have experiences of taking photographs, for example, or video, and storing that on our computer. And a terabyte gets eaten up pretty quickly, when you’re taking high-definition video, for example, home movies. And the kind of data that we’re collecting here at Cedars-Sinai is many thousands of times more than that. Petabytes of data.
Newsroom: Data privacy seems to be in the news every week. What is your department doing to protect patients’ personal data?
Moore: I think it’s a legitimate concern. I certainly worry about my data being exposed. And I think we all do. It’s human nature to want—you know, health care is a very personal thing—and we want that to be private. And we have a lot of good reasons for it being private.
What I can say is that Cedars-Sinai and every other medical center in the country takes privacy very seriously and has a lot of built-in protections for privacy and security, and this is something that every healthcare organization in the country worries about on a minute-to-minute basis, on every day of the year.
One of the things that we’re exploring here at Cedars-Sinai is using a type of AI algorithm to create what’s called synthetic data, where the AI learns the patterns in the patient’s data, but then generates an artificial dataset, which has the same patterns, but the data is completely generated by the computer, so there are no privacy concerns, you can’t re-identify the patients. It’s not actual patient data, it’s completely made up data. But the relationships in the data, the kinds of patterns, the risk factors, and the kinds of things that we look for are preserved, so that computer algorithms can detect that. So that’s something that we’re exploring here, and something that I think could help protect patient privacy when we use patient data for research.
We all take this very seriously, we’re all very concerned about it, we’re all working to protect the privacy of patients and their data.
Newsroom: What are some of the other challenges when it comes to using artificial intelligence and computer algorithms in medical decision-making, and what can be done to overcome those challenges?
Moore: It’s very important to us to develop tools that clinicians and researchers can actually use, that are useable and understandable and explainable and transparent and easy to use, accessible, in addition to being fair and unbiased.
Some of the problems that we’re dealing with come from the healthcare process itself. So for example, we know, nationally, on a national scale here in the United States, that Black patients are treated differently than white patients when they come into the emergency room. So there’s this sense that Black patients can withstand pain better than white patients. And so they might be triaged differently and prioritized differently for treatment. And that is a bias. That is unfortunate, and it hurts Black patients. And so we need to understand where healthcare is biased. And we need to fix those problems at the root. Because what happens is, when patients of different ethnic or racial backgrounds or genders get treated differently, that creates biased data in the electronic health record that the AI algorithms can exploit in unfair ways.
Newsroom: Are algorithms the doctors of the future? Will human decisions ultimately be replaced by computer calculations?
Moore: It’s a question I think a lot of us think about—and maybe some people worry about job security—but my answer to that is that the way I see artificial intelligence is the way I see anybody that we would hire to help out in the clinic or on a research project: that AI is really another expert that is going to look at information, look at data and provide an answer. And that answer can’t stand alone by itself. It needs humans to interact with it, so I don’t see AI as replacing clinicians. I see it as augmenting what clinicians are able to do.