By Patrick Hwu, M.D. - January 21, 2022
Dr. Dana Rollison, chief data officer and associate center director of Data Science, is an expert in epidemiology and data sciences and leads data projects across Moffitt Cancer Center. We recently discussed how her work is unraveling the complexities of cancer through data.
Hwu: Tell us about outcomes in cancer patients specifically.
Rollison: In cancer, a newly diagnosed patient is thinking about life and death. And so survival is something that we take a lot of care to measure very carefully here at Moffitt. Every patient who we treat for cancer, we follow over time to see what their survival experience is with their disease. And we do this in a number of ways. We have an entire department that’s dedicated to recording information about the type of cancer patients have, how advanced the cancer is when it’s diagnosed and then following the patient to determine how long they survive after their treatment. This tracking of outcomes provides information to today’s patients and also helps us figure out how to take care of patients who are diagnosed in the future.
Hwu: What is the significance of publishing patient outcomes? We’re one of the only large cancer centers that publish all of our outcomes. You can see them at Moffitt.org/outcomes.
Rollison: That is correct. We have a team of over 20 people in the Cancer Registry who work to track the outcomes on over hundreds of thousands of cancer patients in our database, over many decades of work that we’ve done here at Moffitt. And we are careful to look at outcomes both one year after diagnosis, as well as five years after diagnosis. And if you go to our website, you can see our rates compared to the national average, by different cancer types, as well as the different stages of cancer diagnosis.
Hwu: How are we using AI algorithms to advance science and improve outcomes at Moffitt?
Rollison: Absolutely. We’re really excited about this department. It’s one of the only Machine Learning departments in a cancer center devoted specifically to preventing and curing cancer. They’re working on a number of very exciting research avenues. As an example, we’re using artificial intelligence to understand the information that’s locked within the imaging of a patient’s tumor, and whether there are certain pieces of data within the radiologic image itself that can be used to determine who will do better with one treatment or another. We also have machine learning algorithms we’re developing for drug discovery. So knowing that a particular drug can target a particular protein that has a certain shape and understanding based on the chemistry, whether there are other proteins that drug might also target is another project we’re working on in Machine Learning. And a third example would be in how we deliver radiation therapy. Our own chair is an expert in radiobiology and understanding how radiation therapy should be tailored to an individual cancer patient. All of these examples are looking at one data type at a time, whether it’s molecular data, imaging, radiation treatment. Ultimately our vision is to be able to combine all of these data types. And it’s very difficult to do that without using machine learning and artificial intelligence.
Hwu: Are there any disease sites that have seen a greater impact from the use of data?
Rollison: Yes, very early on, we had some interesting discoveries from the use of the molecular data collected through our Total Cancer Care program that showed us some patients did better with melanoma if they had a certain gene expression profile associated with immune response. Lung cancer comes to mind as well because of the mutations that we observe in lung and our ability to identify targeted therapies and identify the right subset of patients who will benefit from treatments upfront. We also see survival advantages in lung cancers.
Hwu: What’s the future of data and analytics in cancer care?
Rollison: It’s a great question. I definitely see the imaging being a more automated component of the data gathering. We have expert radiologists who are able to discern whether there is a tumor present and we still need that expertise. The computer can tell us other things about that image that the human eye can’t see, and that information may be important in taking care of the patient. I also see new data types coming about as a result of virtual health and virtual care. We want to be able to deliver our cancer care closer to a patient’s home. The more remote care we can give and remote monitoring that we can do, the better we will be able to deliver care closer to home. That means access to new datatypes from wearables. We have remote thermometers and scales and other types of devices that can send information back to us. So we can monitor how a patient’s doing and that information will give us really more information than we have today because we are able not just to collect information every, let’s say, six months when a patient comes to see us, but we’ll have continuous data to monitor how they’re doing in between visits, which could lead to earlier detection of a recurrence and earlier treatment and improved outcomes. And ultimately we’d love for data sources to connect better. So that interconnectivity of our health care system will be really critical to advance the treatment of cancer patients in the future.