AI and the war against cancer - learnings on tackling "the ultimate Big Data challenge" from Memorial Sloan Kettering

In 2025, the defining chronic condition that impacts on most people’s lives is cancer. In America alone, some 2 million people will be diagnosed this year with some variant for the first time. In fact, it’s estimated that around 40% of all Americans will have a cancer diagnosis at some point in their lives - half of men and a third of all women.

So the battle against cancer is very much on the frontline of healthcare. In 1971, the National Cancer Act in the US came into existence to create a national response to the disease and how to tackle it. Among the institutions leading the fight is New York-based Memorial Sloan Kettering Cancer Center (MSK), one of 72 National Cancer Institute–designated Comprehensive Cancer Centers.

Chief Strategy Officer Dr. Anaeze Offodile notes that the National Cancer Act declared “a war on cancer”, and while the statistics cited above are gloomy, he does insist:

I'll say we've come a long way since then. We understand what's driving this prevalence. It's genetics, it's lifestyle, it's environmental exposure. I'll say the biggest insight that we've gotten is no cancer is the same for the same diagnosis. It tends to behave differently.

Data problem 

Those inherent differences add to the complexity of treating the condition. They also lead to Offodile making a simple declaration:

I think cancer is the ultimate Big Data challenge.

He expands on this thesis by pointing to all the various types of data that just one cancer patient will generate on a daily basis as healthcare professionals and institutions engage with them on their treatment program:

Now multiply that by the million patients we engage with in a given year at Sloan Kettering, that's a lot of data.

In fact around 90% of cancer data on record at present has been generated over the past five years, he suggests, so the rate of growth of the amount of information that has to be handled is growing at a rapid rate. And, of course, it’s being stored and processed in systems that have not been built to cope. Offodile says:

There are a lot of silos in healthcare. It's typically unstructured, purpose-built, relational data. So it's key that we have a technology partner…to help us make it secure, ensure the patient privacy, make it usable, make it standardized, and make it integrated.

But that in itself, while commendable, is far from the end of the story, he adds:

If you do that, that's necessary, but not sufficient, because there's just so much data coming at us every single day that no one person can make sense of all of it. That sense-making is where AI comes in, and whether it's agentic models, NLP (Natural Language Processing)  models and other lands, we're able across use cases about how we can use this data to best serve our patients, for personalized therapies, personalized recommendations, to get the best outcomes for our patients. It's a data issue, it's a technology issue, and that's what we are today.

That being the reality of the situation, MSK has partnered with a number of tech companies, most recently with Amazon Web Services in an initiative designed to help develop cancer patient treatments and track cancer research. According to the formal announcement of the tie-up back in February:

The goal will be to build a high-quality, up-to-date longitudinal data resource for cancer research at MSK, as well as serve as a source for validating and improving cancer research with partners. With this, MSK researchers and providers can better track how a patient’s cancer changes over time using computable disease trajectories. Combined with advanced tumor and clinical response predictions, this data-driven approach can help uncover new insights, personalize treatments, and improve patient care.

AI potentials

There are, obviously, a number of clear applications for AI tech, such as examining patient data for predictive purposes to allow for early interventions and preventative courses of action, as well as being able to present a range of treatment options and their likely efficacy. AI can also, in theory, go beyond what humans can in certain areas, such as detecting anomalies and patterns in data that might otherwise go unnoticed.

Of course in patient care, as in all other applications in other areas, AI is not a silver bullet, however evangelical some of its more enthusiastic disciples are. For his part, Offodile has been pragmatic in his assessment of AI’s impact, stating, for example, to Yahoo Finance:

There's a lot of attention, a lot of hype, on AI, but I'll say it's still a nascent field. There's been no clear transformational or canonical discovery in AI, yet. Lots of potential. Lots of stars being born and funded. With humility, we certainly don't want to overstate what we can accomplish, but I think we have maximized the odds of success by having these two enterprises thoughtfully engage.

But there are wins to be had. Offodile cites the example of a patient, Simone, who comes to MSK, gets a biopsy, followed by a genomic driver of her cancer:

Based on that, we can now create personalized therapies based on her history, her lifestyle and the genomic profile. And if the next best thing is surgery, we're now able to connect her in to surgery. Even within that very technical undertaking, we can have AI through computer vision, make her robotic surgery be as seamless as possible, and allow the surgeon to only take out all that he or she needs to clear the cancer. That's key, because sparing as much tissue as possible is a better quality of life, less time in the hospital, and maybe not needing oxygen. So that's a really, really important outcome - to do personalized AI supported surgery. After surgery, you may learn that she has a clinical trial that AI can match her to the trial, or that her her genomic driver, has a special target therapy that can give her better cancer cure.

That’s the promise and potential of AI, he argues - to deliver precision medicine that is scalable. He concludes:

Through Simone,  we can learn how others have fared and update all those learnings with her own personal journey, and make that available to patients who come through our doors every single day. So at MSK, our mission is to end cancer for life. It's why we exist. It's what binds our people, our science, our culture. All three aspects of that can be and are and will be optimized with technology.