Can AI Help Us Solve Cancer?

Forgive me for being a bit cynical

Me, leaving A.I. behind (Painting by author)

When my daughter Ana was diagnosed with cancer in 2012, I was working as a self-employed search marketer. I had clients, a fixed daily routine, and a very bright future. My girls were eight and eleven. They were the center of everything. Life was good until our luck ran out.

Ana’s diagnosis was a whirlwind cinematic moment of terror — a bad stomach ache, a trip to our local hospital, a white-faced physician explaining the horrific results of her CT scan, an endless ambulance ride to a children’s hospital, and our first midnight meet and greet with a pediatric oncologist.

The hospital stay would be indefinite, so I emailed my clients to let them know I’d be MIA for a while. I apologized. I said I’d try to work at her bedside. I delegated the most important work to my subcontractors.

One of my clients (Dr. N), a spine surgeon, immediately responded. He told me not to worry about work. It was more important for me to be present for my daughter, be her advocate, and get at least 6 hours of sleep a night.

He said he had seen parents completely fall apart in hospital scenarios like the one I was about to face, mainly due to lack of sleep. Over the next few weeks, he provided additional guidance about how to navigate the complex hellscape we’d found ourselves in.

He was an expert in the world of hospitals. I was merely a visitor — confused, exhausted, perpetually filled with dread. In the fog of those earliest days of Ana’s diagnosis, Dr. N helped me understand how important it was to be her advocate. He also gave me hope.

Medicine is a waiting game. Dr. N explained that when something rare and terrible occurs, something with no clear path to treatment, the key is keeping the patient alive as long as possible.

“We need to buy time,” he’d said, “because there’s always another drug or treatment in the pipeline. Get her to the people who are experts in her specific type of cancer. That’s the best way to make sure she survives long enough for a miracle.”

I don’t think these were his exact words. It’s what I remember fourteen years after Ana’s diagnosis and nine years after her death.

I didn’t think Ana was going to die. But just weeks after my indoctrination into Cancerland, I was already starting to understand the limits of modern medicine. There was exactly one treatment for Ana’s rare tumor — surgical removal. Since her tumor was the size of a grapefruit and it engulfed her liver, she needed a liver transplant.

It was supposed to be the cure. Her oncologist was very reassuring about this. Remove the tumor and the liver along with it, and she would recover. It was a terrifying surgery for an 11-year-old to face, we had no choice, no other options. Six months after that frantic ambulance ride to the children’s hospital, Ana got her liver transplant. Six months after the transplant, her cancer returned.

Cancer recurrence is an outcome we dreaded — going back into the hospital, facing constant scans, putting Ana through the torment of brutal treatment right when she was starting 7th grade. It felt impossible. But I recalled my client’s words and was still quite hopeful. We just needed to keep Ana alive long enough for someone to produce a miracle.

And that’s exactly what we did. At least, for a while. When I say we, I mean her brilliant oncologist and the surgeons who kept removing tumors as they popped up throughout the course of Ana’s illness.

Ana had inflammatory myofibroblastic tumor (IMT), a rare presentation of a rare cancer. This is exactly the kind of cancer that compels doctors and scientists to push the boundaries of medicine.

Ana’s doctor sent one of her tumors out for a complete genomic sequencing, a process that maps the DNA and RNA of a tumor to identify its specific genetic mutations. This is a relatively new technology. Widespread mapping of various cancers wasn’t initiated until late 2005 when the NIH launched the Human Cancer Genome Project (HCGP).

Among other things, genomic sequencing is used to help doctors identify targeted therapies based on gene mutations that the sequencing identifies. When we got the results of her sequencing back, they weren’t what any of the brilliant minds at Columbia Presbyterian expected.

Here’s how Ana’s doctor explained it (I kept the notes): “There’s a fusion of two genes that’s very novel — meaning, we’ve never seen it before. This fusion involves a receptor called IL23 which activates certain pathways involved with inflammation that cause the gene to be continually on. There may be a drug that targets IL23.”

Incredible news. The drug he identified was called ruxolitinib — brand name, Jakafi. It was approved on November 16, 2011. That meant it had only been available for about three years at the time Ana started taking it. The drug was approved for something entirely different than Ana’s cancer. It had only been used in adults.

By the time we got the call about her specific novel cancer mutation, testing with Jakafi had begun on children. Data on the results of the test had been published in an oncology journal just a couple of weeks before Ana’s results came back. Ana’s doctor reached out to the author of that study to get contacts for the company that made the drug. And, miraculously, he was able to get it for her. It was even covered by insurance which was a good thing because Jakafi cost about $8,000 for a 30 day supply.

We’d done it. We’d kept Ana alive long enough to find the miracle treatment that could potentially save her.

And it was a real miracle that the novel mutation was discovered and that Ana’s brilliant oncologist was able to identify a drug to treat her cancer. Jakafi worked, for a while, but cancer is sneaky. It often mutates and becomes resistant to treatment.

Ana was on several different targeted drugs at the end of her life. These medications kind of worked, but they never completely stopped the growth or spread of her cancer. I got Google alerts for IMT, and when a new drug was mentioned, I forwarded the study to her doctor. That’s how we lived for the last few years of her illness, trying to outrun Ana’s cancer with every treatment available to her at the time.

And then time ran out. The drugs stopped working, the cancer kept spreading, and Ana’s story ended.

I’ve been thinking about this experience a lot lately, given the conversations around AI and its role in medicine. Getting targeted drugs to the patients who need them isn’t easy. The drug discovery process is long, though new technologies like Google DeepMind’s AlphaFold, are using AI to speed up the process.

AlphaFold predicts a protein’s complex 3D shape directly from its genetic sequence. It can map these structures in minutes, a process that once took years. This helps scientists design targeted drugs much faster.

Demis Hassabis and Dr. John Jumper at Google DeepMind won the 2024 Nobel Prize in Chemistry for the work they did on AlphaFold. Now Hassabis also heads up Isomorphic Labs, an Alphabet sister company focused on drug discovery. This marries the two halves of what’s needed to “solve all disease” as Hassabis has said — identifying proteins and matching them to the drugs that will treat them.

And this is also one of the key rationales that the biggest players in AI use to justify the relentless push towards AI everywhere, all the time, all at once. AI will cure cancer. AI will solve hunger. AI will make everything better. In every interview I’ve watched with Hassabis, and every article I’ve read about the topic of AI and cancer, it seems he is sincere in achieving this goal. But even if the technology suddenly produces miracle drugs for all the things, will it be accessible to patients? Will it be worth the collateral damage it creates?

I feel bitter. I recognize that we need noble pursuits like curing cancer. We need research into rare diseases that don’t necessarily equal profitability for big pharma and biotech companies, but can save kids like my daughter.

But it seems to me that we may pay a huge price for this lofty goal, and that no one, at least in the short term, is going to be saved by allowing AI to push forward with no reins or guardrails or real plan. And while all this money is being thrown at AI, the U.S. has cut funding for all kinds of science research, freezing or terminating nearly 8000 research grants since January 2024. They’ve also cut 25,000 scientists and personnel from the agencies that oversee this research. The largest grants cut? Those funding clinical trials needed for drug approval.

AI can help us get the drugs faster, but the drugs don’t reach the patients unless we have clinical trials. There could be a drug out there that can save you or your child, but so what? It may as well not exist if it can’t get approval, isn’t covered insurance, or isn’t tested on, and ultimately approved, for children.

I wonder who is doing the good work here. Is it Hassabis and his ilk, pouring billions into AI with the feverish conviction that it will one day solve all of humanity’s problems? Or is it the overworked pediatric oncologist and scientists who figured out how to get the right drug to the right patient at the right time? Are these people even talking to each other?

Medicine is about buying time, but it’s also about connecting the dots. The massive amount of money being fed into AI systems will ultimately be money wasted if there’s no investment in the other side of the medical coin — the human side. It won’t matter if AlphaFold identifies the perfect drug without a way to get that drug to patients. We need these miracles, but I keep wondering if guys like Hassabis really understand what’s at stake. Is he using “solving cancer” as an excuse to tap into Google’s bottomless resources and realize his own AI dreams?

You’ll have to forgive me for being a bit cynical for wondering if the goal isn’t to save children like my daughter, but to win this guy another Peace Prize and to enrich companies that are already incredibly wealthy. I had a glimpse of what a real miracle really looks like, after all. I know that “curing cancer” sounds great in theory, but the reality for many families like mine is that there just wasn’t enough time for a true miracle to save our kids.