You heard a little about it at the very highest level in the video we just saw, but how we’ve gone about it in partnership with the Watson Research team. So, as Ginni just said, this is a machine approach that provides a cognitive approach to understanding complex problems.
So, how do we go through the cognitive training? That started with a collaboration between the two research teams, two years ago, as Ginni said. We put our best physicians, of the thousands of physicians we have taking care of every year 125,000 patients with cancer, all the experience built into those individuals, and went through a training process of the cognitive abilities of Watson with the research team.
We have refined that over two years, you met the physician leader in that video, Mark Kris going forward. Out of that two-year process, we have reached a point where we have developed a partner for the health care professional in making the best and most informed decisions for patients and their families about the diagnosis of cancer.
How does Watson actually do this? In the largest overview, there are really three parts of the process. Watson has learned how to retrieve all the relevant information that’s necessary for a personal decision of that patient and their specific circumstances relative to their cancer. All the medical information that exists in the world as well as all the personal information that’s in that person’s health care record.
Second, once it’s retrieved, it can use its cognitive ability to integrate all that information. And in the way of the world, often patients worry that someone else is making the decision. Watson is a collaborative partner with the health care professionals, because from that information and that integration of cognitive abilities, it provides a prioritized list of what are the best possible choices of among the full range of choices for the best diagnosis and treatment of that patient’s illness. And it works in partnership, as a collaborator, with the health care professionals that are dealing with that patient and their family about their treatment. That’s the overview.
I’m very pleased now to introduce our physician in chief, who is in charge of all thousand of those health care professionals that have been training Watson from the medical perspective to give the specific overview – Dr. Jose Baselga.
José Baselga – Physician-in-Chief of Memorial Sloan Kettering
So, good morning everybody. And this is a standing room only, gee, you are all the way in the back so I hope you can see the slides well. I’d like to present some practical aspects on the challenges that we are facing today in taking care of patients with cancer and then it all will click together. And I hope I can demonstrate clearly that Watson will become a critical partner in the way we deliver care to thousands of patients not only at Memorial Sloan-Kettering but all around the country and hopefully around the world. This is our traditional oncology decision making process. This is what Dr. Thompson learned when he was a junior faculty; that is what I learned when I was a junior faculty; and, this is the tradition of centuries of practicing medicine.
Our process to make decisions in cancer care was based on looking at the chart of the patient, and it was a paper chart with limited information. Looking at some x-rays, very basic x-rays. You know, things got complicated and then we had CAT scans, but we had very few CAT scans, and that was it, what we had. Some laboratory data. We had a few drugs, not too many.
When I trained in breast cancer we had one hormonal therapy and three types of chemotherapy, that was it. So, our decision making process was in this way simplified. And then of course, we had books at that time. Actually, in oncology, we had “the” book that we had in our clinics and that we would consult often, but it was just a book. And the number of papers that were published every year was very minimal and there was no Internet. So, that was something that we felt comfortable in doing.
We had these data that was being analyzed in the clinic and then we offered to our patients our best decision and our best proposal for therapy. Fast forward, this is not any longer what’s happening today. The field of medicine has changed in a way that we could never have predicted. And there’s a sea of data — like Mike has mentioned to you — that we have to deal with in every single aspect of our patient care delivery.
To begin with, who is reading books any longer today? We have thousands of articles that come up every day to address every single issue of our patients. Every single need, every single question, it’s written in thousands of articles that come out. We have electronic medical records with tremendous amount of data about the patients: their prior therapy, family history, response to therapy, et cetera, allergies, interactions.
An imaging revolution: we are talking about CAT scans, we are talking about PET scans, we are talking about sonograms, we are talking about multiple imaging techniques that on their own are extremely complex to interpret and to understand and to integrate into our clinic space.
Therapies. Well, in breast cancer now, we have today 80 therapies, but we have 800 therapies that are being studied in clinical trials. Eight hundred therapies — from four to 800 in 20 years. And then of course, what was being referred to, we are just not looking at the pathology slides any longer; we are sequencing our tumors, and we are analyzing 20,000 genes and we will analyze even more complex parameters in every tumor. And our medical knowledge at Memorial Sloan-Kettering, from a few hundred physicians to a thousand physicians that we have right now. So, a tremendous amount of knowledge that is in our system.
So, this is the rise in medical publications, exponential, and this is not going to get easier. The need to develop physician-based medicine based on genomic data — every patient’s tumor is different, every tumor is different. We don’t have just two types of breast cancer; we have now at least 50 types of breast cancer, and this happens to every tumor type. And we will need to have a matching of the patient’s tumor characteristics with the specific therapies and be able to monitor this very carefully.
And just to give you an example, this is what’s happening today at Memorial Sloan-Kettering Cancer Center. We are routinely sequencing the tumors of thousands of patients by these new platforms that I will not go into in detail. But just to give you the sense of complexity, we can genotype genes in full. And this is our list. And you can read the list, and this is done on purpose. You’re not supposed to read the list, because even if you read the list, you will not understand what it means and nor will a physician understand what it means. And every one of these genes can have multiple mutations. And these mutations, among them, interact. You cannot deal with this even if you have an hour per patient in the clinic, which is not the case, by the way. So, the traditional process is not working any longer, is not working, and that is where Watson comes to play, and that’s why we are so terribly excited.