Despite all the talks about whether artificial intelligence algorithms will be replaced by doctors, Eric Topol is not worried. Topol is a cardiologist at Scripps Research Institute, a geneticist, and author of several books about future health care.
His newest book is called Deep Medicine: How to Make Artificial Intelligence Health Care Human Again (out today from Basic Books). Topol said people are always excited about bonds that other people take care of and that the AI can help improve that bond and return it – if doctors want to stand in business interests.
Verge spoke to Topol about how today's health care works, health-related health concerns in the AI age, and the importance of physician's activism.
This interview was edited slightly for clarity.
Before we talk about how AI affects healthcare and patient-doctor relations, can you tell me what the relationship looks like now?
The relationship has worsened. It's really awful. The patient is short-lived because you get very little time with your doctor and also do not interact with you for a little while. And not just hours. This is the excitement of doctors who serve as data clerk. You will not be a good listener in that case. Doctors are extremely disenchanted and disillusioned and burned and depressed.
Plus, we have all this data – collected from genomes and sensors – and it's all dressed in no way. I opened the book with one of the stories that fortified me, where this information was not being processed by doctors and people who cared for me, and I was injured and my recovery. Being reinforced as if all of this data is teed up for people, we can make things safer and have better results and have things become better, but doctors too busy. They can not get their weapons around the data of every person. I think the potential transformative of AI is in its power to enhance the aspect of human medicine, which is something we have lost.
The aspect of human medicine is why you do not think I can change doctors, right?
We do not just need human management because you do not always trust an algorithm even if it's proven – it may be hacked or glitches – but I think we always have a quest for that bond, intimacy. We use it. This is important, and I remember that. What happened over time took the medical business, and all these forces fell into that relationship. We can get it back.
Let's talk about the diagnosis of AI first. As per week, there is a study on how the AI can assess some of the conditions better than doctors. How to play it?
AI can see things people can not do. Intensive learning machines begin to see things better than what a person can see, and we begin to realize all these things that we never guessed. There are many examples today. You can determine the potassium in your blood at your watch without any blood. You can check the retina to see if it is male or female with high accuracy. You can study a colonoscopy, and the vision of the machine will get polyps that GI doctors did not get. The list goes on and on.
The missing piece, of course, is a careful, rigorous, prospective study of validation and replication. We have a promise today. We have seen enough data and it is exciting as anything I have seen in my four decades in medicine, but we also need to take it from the disturbance and hyperbole to the level of truth and there is no apparent evidence.
You see a world where we have diagnostic help and also algorithms that incorporate all the data resources together. What will this be in the clinic?
Then we have another world. When you see patients, you are not trying to work your way through all the different pages and sources of data. You are in, "Okay, I will have a context for my patient. I will have a meaningful relationship and understand the existence of a person to give my human wisdom and empathy." This is a different look than what we have today.
Everyone benefits if you get better and the doctors stand up and say, "We'll give this benefit back to our patients." People are getting more and more charged with their data and support through algorithms while at the same time, they eliminate the burden of clinicians. And then the clinics get their performance and efficiency, and they remember why they did the medicine in the first place. It creates a flywheel effect. You get both sides who are getting this performance improvement, and it changes the whole perspective for clinicians.
What about privacy?
That's really important. It is important that each person owns their data and combines them all so that it is not just what is in your medical records but is scattered in many places and hospitals and sensors. Right now, no one has all of their data, even if you want it from the moment you are in the womb when you get an assessment.
The biggest thing we can do is to own data to people. We need to examine data security and privacy, but it also involves a different model of ownership.
How do you know that excellence is restored to patients, rather than force doctors to see more patients in shorter periods?
I spent several years commissioned by the UK government to help evaluate and study the National Health System. We have economists working on them, and it is noteworthy that every minute you save in voice recognition environments [where doctors aren’t sitting at a keyboard inputting information] translates into an enormous amount of time to free for doctors. The multiplication of effects is quite amazing.
AI can restore the efficiency and productivity and workflow, but if we go to this route, we must have the will to stand up for our patients. That has not happened in the past. If we continue to live as we have, the medical community is more restrictive, and there will be more burnout and more depression and suicide. The real test is if the medical community can stand on financial business interests. Sometimes we are unconditional, and we can not afford it again. It takes activism to make people more humane as machines become better and improve the effects of humans.
What types of activism can doctors do?
You have not used to see the doctors standing. In recent years you have seen the National Rifle Association that says "stay on your lane" in the gun policy, and you see the doctors stand. These tend to become younger people, not old dogs, which are too much of us in medicine. You begin to see the speaking physicians, and we can do that in a large amount for the most important thing everyone has, which is the restoration of health care care. I am confident there is one way to do that.
How far is it all?
We have the evidence that people are interested. It's moving fast, but over the years, I've learned that whatever I estimate for how long it should take, maybe I should multiply by four or five. I learned that even though you have something exciting about it, it lasts much longer.
There is a lack of investment in producing high-quality research required. A lot of the best jobs in this space are startups that form a radiology algorithm or a dermatology algorithm or clinical voice recognition. They do not have to have resources, unless they get a Google or an Amazon (without a track record of doing rigorous medical research either).
It's hard to show proof. The medical community and, for those things, patients, are not likely to accept reconfigured health care without proof. Of course, it should be rigorous because if you have a broken algorithm, you can hurt many people who are really fast so there should be very strict standards and research requirements on large numbers of diverse people and different place. We need proof that nobody can argue, and then move it faster, and start getting the momentum we need.