How AI Could Help Doctors Across the Cancer Care Team

Artificial intelligence is increasingly being used in cancer care, but its potential role may extend beyond oncologists and cancer specialists.

Patients with cancer often have other health conditions, take multiple medications and see several physicians throughout their care. As a result, doctors outside of oncology may also be asked to weigh in on complicated cancer-related questions.

Dr. Maurie Markman, professor of medical oncology and therapeutics research at City of Hope Comprehensive Cancer Center and president of Medicine and Science for City of Hope, discussed how AI could help physicians across specialties navigate those questions while keeping the doctor at the center of patient care.

CURE: Cancer patients often have multiple doctors involved in their care. Why could AI be particularly useful for physicians who aren't oncologists?

Markman: I think that's fascinating. If you look at the data, cancer is the number one disease entity that physicians are asking about, and a significant portion of the questions are related to therapy.

What's fascinating is that the majority of the doctors asking these questions are not oncologists. They are family doctors, general internists and other physicians.

That's not surprising, because who sees these patients? Yes, they may see an oncologist, but they have other diseases. They have primary care doctors. They have heart disease. They have diabetes. They have all these other issues that need to be considered.

So you have primary care doctors who may be asked very complex questions about cancer by their patients or their patients' families.

Why might a patient turn to their primary care physician with questions about their cancer treatment?

Markman: Patients trust their primary care doctors. They may have known them for 25, 35 or 45 years.

Maybe they don't have the same relationship with their oncologist, who they've known for six months and who has essentially told them, "You have a bad disease."

The patient may say to their primary care doctor, "What do you think? My oncologist told me this. What do you think?"

The primary care doctor might not know the answer. But they could potentially use AI to look at the patient's condition and help provide information.

That doesn't mean the AI is making the decision. It's helping that physician become more informed when they're trying to help their patient.

How could AI help when a patient has cancer as well as another health condition, such as diabetes or heart disease?

Markman: That's where it becomes particularly complicated.

You could have a 45-year-old patient with a particular type of cancer who also has an underlying endocrine disorder and a blood sugar of 195. There are thousands of things you might need to think about.

The patient may also be taking multiple medications that have nothing to do with their cancer. What about the interactions? What about the possible best therapy? What might be the best sequencing of therapeutics?

The idea is that you could take all of this information — FDA approvals, clinical trial data, the medical literature and expert opinion — and ask a question.

AI could potentially help pull that information together.

Does that mean AI could help different specialists communicate more effectively about a patient's cancer care?

Markman: That's one of the possibilities.

Expert opinion doesn't necessarily mean just the oncologist. How about the endocrinologist, the cardiologist or the nephrologist who can help with the various organ systems that are already at risk because of underlying disease, but also because of the toxicity of the drugs?

How do you mold all of that information together with the question you're asking?

That's where I think AI can be helpful. It can help bring together information that a physician may not otherwise be able to recall or find quickly.

But again, it isn't taking over. It's providing information and insight.

You've described AI as something like a "curbside opinion." What does that mean in the context of cancer care?

Markman: In my era, the concept of a curbside opinion was literally that you'd talk to a colleague in the next examining room who was seeing a patient.

You might talk to a colleague at lunch or breakfast, or even as you were both going home at 9 o'clock at night.

You'd say, "Hey, what do you think about this particular patient?"

The idea is they would throw something out to you. You could ask them, "That's interesting. What do you think?"

It could be a back and forth. It doesn't give you the answers. It helps.

AI can potentially serve in that role — as a companion, as a second opinion or as a source of input and insight.

What should physicians keep in mind when using AI in that way?

Markman: The doctor has to remain the person making the decision.

I use the term "cognitive surrender." That means essentially saying, "Hey, let AI do it for you."

No. Never.

I can't imagine that.

The doctor is the person the patient and family trust. The doctor also brings their knowledge and experience to the situation.

You have to look the patient in the eyes and answer the question.

AI can be a great companion. You can ask, "What do you think?" Or, "Did you consider that?" Or, "Did you remember that the patient had a myocardial infarction a year ago?"

Maybe you didn't remember that. How could you possibly remember everything?

But AI might prompt you to think about something. It might make you consider germline testing. It might make you think about renal function. It might make you think about asking an endocrinologist.

It provides input and insight. It does not take over for you.

It makes you a better doctor.

Could this type of technology be especially helpful as cancer treatment continues to become more complicated?

Markman: Absolutely.

Cancer is incredibly complex. We're not talking about hundreds of different diseases. We're talking about thousands of different diseases when you consider the different mutations and mechanisms involved.

And the therapeutics have become almost as complicated as the biology.

The rate at which new therapeutics are coming along is extraordinary. Then you have clinical trials, treatment guidelines and all the information being published.

It's impossible for any oncologist, no matter how specialized they are, to know everything.

That's why I think AI is not only important but essential as we move forward.

What could AI's role look like for patients who are navigating cancer information themselves?

Markman: I would hope that in the future this could become helpful in answering questions for the public as well.

I'm talking about reliable platforms.

We have a pandemic of misinformation related to cancer. Patients can go on social media and find somebody with hundreds of thousands of followers, but that doesn't necessarily mean the information they're receiving is accurate.

If we can develop reliable AI platforms, I could see them becoming an incredible form of education.

They could help people find information about things like cancer treatment and even vaccinations, including HPV vaccination.

The potential is there to use these platforms in a very positive way.

What concerns you about AI becoming more widely used in cancer care?

Markman: I think the biggest question is how do you assure quality?

How do you make sure these platforms remain excellent?

I'm not necessarily talking about a malicious actor. I'm concerned about what happens if these platforms become proprietary and companies have financial relationships with them.

For example, what happens if drug company A or drug company B pays a certain amount to the platform?

I think it's fair to worry about those kinds of things.

We need to make sure that the platforms remain focused on providing high-quality information.

Ultimately, what do you think the relationship between AI, physicians and patients should look like?

Markman: I think this is an incredible development, and I think it's essential.

Obviously, there needs to be a lot more thought put into how we move forward and how we make sure it continues to be of outstanding quality for all our patients and for society as a whole.

The physician needs to remain at the center.

AI can provide information. It can provide insight. It can help you think about things you may not have considered.

But ultimately, the doctor is the one who knows the patient, understands the context and has the responsibility to make the decision.

For more news on cancer updates, research and education, don't forget to subscribe to CURE®'s newsletters here.