Brantley Thrasher considers the future of prostate cancer and the approach to treatment he expects to see in the not so distant future.
That’s a complex question, but one of the parts that I think we will see – that will transition the practice of urology – will be artificial intelligence (AI), deep machine learning and large datasets. I see that being used already in practices.
So for things like imaging technology and tissue recognition for pathologists, and specifically for radiologists, you know now that we can teach the machine how to look for metastatic disease, for instance. I think that’s going to change the way we practise – using artificial intelligence, and these very fast computers, I believe to start looking at risk factors for prostate cancer, for instance, that is being talked about quite frequently.
Right here in Australia, there was a study that looked at multiparametric MRI, and how the machine can actually be more accurate at diagnosing an abnormality on the MRI; they’re potentially a radiologist. So I think it’s an exciting time for us, I think we’re going to be using this big data, big datasets, to help the patient and a physician.
How is this going to help the patient?
I think patients have so many potential data points. You think about things like prostate cancer – you’ve got family history, you’ve got the race of the patient. Consider, for instance, how many of the potential genetic abnormalities they may have, BRCA abnormalities – the BRCA gene. They have a lot of things that we can compute, but for me to sit down and do it on paper is impossible.
For a machine to do it very quickly, and to get the answer to the patient: ‘This is your risk, this is why you should probably undergo a biopsy.’ It helps us with that risk assessment, and helps us to tell the patient a potential risk versus the benefit. That’s going to be a lot better for the patient to make that decision. Do I undergo a biopsy, yes or no? Do I have a cancer, or what’s my risk for cancer? That allows me to better tell them what is the risk/benefit ratio in their favour.
It’s very difficult when the patient walks in to say, ‘Because you’ve got a father that had prostate cancer, you’ve got a brother that had prostate cancer, that increases your risk four-fold.’ Then you start adding in the other things and you start talking about genetic abnormalities that we don’t really know. The machine can take all of that data and spit out a risk assessment so much better than we can.
Clearly, you’re a supporter of the technology?
Absolutely. Used correctly and under certain security protocols, no doubt.
What’s going to be the biggest area of change, do you think?
Oh, big data – there’s no question. We have so much data out there right now. I think that if you look at the millions of data points that are there, we have the genomic profile for patients, so we have so many particular genes that can put you at risk. Having a particular gene, for instance, putting you at risk for particular cancers, for heart disease, chronic abnormalities like diabetes.
When we look at the genomics, proteomics, all of that data put together, with ultra-fast computing, it gives you a much better idea of what you’re at risk for having, or being diagnosed with, and whether you should seek attention or medical advisement early on. What’s going to be critical to the change in practice is that we’re going to have these datasets that can be crunched. It’s going to help us and them decide: ‘Do I need to get more done for this colon cancer, prostate cancer?’… whatever. Even heart disease and diabetes, chronic diseases that we deal with a lot.
The gathering of this data on particular profiles, DNA… is this a good thing for patients, or should they be wary?
I’m a little worried. I’ll be truthful; I think that there are concerns about security. Everyone is worried about being hacked. I think there are concerns on how this information will be used by third parties. Is it possible you could be denied insurance? I think that’s a potential risk. Why? Right now, if they know that I jump out of planes as a hobby, or I fly a plane as a hobby, my insurance premiums go up. If they know, for instance, my house is in a flood plane, my insurance premiums go up. What’s to say that they wouldn’t do the exact same thing if they know what’s going on with your genes?
Our current government’s talking about My Health and having all of Australians’ medical records put into one place, which could be accessible only by the patient, with their permission. What are your thoughts on that?
I think it’s great for the practitioners especially, because the crosstalk is always a worry. We have so many silos in the States, we have people that have particular datasets, we have different electronic medical records, we have Epic, we have UroChart; we have so many different datasets that aren’t talking to each other.
Even in pharmacies, it makes it impossible. The opioid epidemic – we don’t know that a patient got this at one pharmacy and refilled another opioid prescription at another. So they’re out there shopping around, there’s not a lot of crosstalk. But when you have it in one place I think it’s impossible for you to think, ‘Well, it’s always going to be secure.’
They don’t know by whom, but they do know that the military database in the Department of Defence was hacked. So to think that it’s always going to be secure is a significant concern.
Will we get to the point where diagnosis and treatment is totally robotic?
I don’t think it’ll be totally robotic. My personal feeling is that robotics, deep machine learning, [IBM computer system] Watson for instance, can help us on a smartphone. It can diagnose a patient, for instance, because Watson’s learning an awful lot through the Memorial Sloan Kettering Cancer Centre right now, about cancers.
They’re feeding all this information into Watson and ultimately I should be able to take my phone, pull up Watson and say, ‘I have this very rare cancer right now, and this is what I’d like to know: what’s the best clinical trial in the world? This patient should be a person that would meet those clinical criteria.’ Watson should be able to spit that out to me.
Interestingly, I think it’ll be a great adjunct. Do I believe that robotics, machines, will take the place of physicians, general practitioners, urologists or specialists? I don’t. I don’t think that patients are going to allow that. I don’t think that you’re going to sit down with a robot and feel that this person is empathetic; they can hold my hand and make me feel like I’m the most important person in the world. I’m going to treat them like a family member. If we can teach a robot to do that, I’m worried. I don’t think we can. They’ll be adjuncts, not replacements.
You have spoken about the future of urology, where patients may be looked after in their homes rather than in the hospital. Is that a reality?
I think so. A lot of biosensors are being developed – a contact lens, for instance, that’s going to, on a real-time basis, measure the serum glucose levels for patients, rather than having to have blood sticks. They have wearable and implantable devices now that will do things like check through genomics for sexually transmitted diseases or urinary tract infections.
Think about a lot of these people in assisted living – you get those folks, put them in an ambulance and take them down to the hospital or to a clinic. Wouldn’t it be great if with these biosensors – some of those are biodegradable, some are not – you could let them check their urine right there at home? Discharge your patient earlier by a day, and let them be monitored at home. You can measure their temperature, possible urinary tract infection, a variety of things, like vital signs, by these biosensors.
I think it’s going to be done because the most expensive care you can give is in a hospital. So, if you can get them out and let them stay in their own home, we know they prefer that, we know that it’s less costly. If we can do that by sensing again, with a lot of these biosensors, wearable devices, I think that’s the way of the future.
Regarding DNA and future privacy issues, do you think the laws and regulations are up-to-date with where the technology is at?
The answer is no. I think that there are a lot of potential places where this could be a problem. I’m going to give you an example, and this is my own personal opinion. You have places where you can send in your DNA and that DNA is used to see what your ancestry may be.
Now think about that for a minute. You just paid to find out if you may be Scottish or English, or what’s your background. ‘I thought I was Italian; well, I’m not.’ So when you send that in you’ve basically signed over a lot of personal information, as well as your DNA. There are millions of those specimens. Don’t you know that an insurance company would absolutely love to have their hands on five million people like that? To be able to say, I know if this person has a very risky gene, we’re not going to be able to insure that person.
So from a legal standpoint that’s going to have to be regulated. If not, I think there’ll be a lot of people that could potentially not be insured, or be put at higher risk or considered higher risk, and we don’t know. We say, ‘Well, this is a possibility, they’ve got a 75 percent chance of developing this, they’ve got a 25 percent chance of not.’
So could that be something that we need to regulate? I think the answer is absolutely. I think that third-party payers would pay millions to get their hands on data like that.
We have a robotics training facility opening in Victoria next year. What are your thoughts on this?
Training surgeons just like the aviation industry, with simulators and virtual reality, I absolutely think that’s the way with the future. I’ll tell you why I feel that way. First of all, I’m a pilot. Second, I do primarily robotics, so I see the similarities and I spend time in an aviation simulator (I’m staying current on my instruments).
Think about it for a moment: dentists have to fill a tooth or put on a crown to be able to graduate from dental school, pilots are staying up-to-date every six months, spending time in simulation, with a lot of things that are very important. They’ve got an engine out, smoke in the cockpit, loss of pressurisation and lives are at risk. Well, it’s the same thing with surgeons.
How do we really test these individuals? Right now we have [an attendee] watching them do an operation. In my case, all I watch them do is prostatectomies. I have no idea how good they are with the robot with kidneys or bladders, or anything else. So it becomes a difficult situation if you can’t watch them actually do surgery in a variety of emergency situations. It’s costly, but I think it’s coming.
I think we’re behind our aviation partners in other places, but I do believe it’s going to be an easier way for us to be able to test without putting people at risk, meaning the patients. We’ll be able to test skills, and put them in emergency situations to see how they respond, and that may be used for credentialling. It certainly could be used in the future for certification.
I think there’s no question that’s where we’re headed. How quickly? That’s going to depend on the technology to some extent, as well as how well we incorporate that technology.
A lot of the repetitive tasks that we do today – things like drawing blood, sterilising the operating room – are things where I think robots are going immediately. They’re already starting to be incorporated into our practice.
Xenex [germ-zapping robot that uses UV flash lamps] and you can sterilise… Think about it, it takes an hour sometimes, the turnover on an operating room. If you could push a flash lamp in there… it’s being used by three hospitals in the US right now.
Then the Veebot [which automates needle insertion procedures]… how many times have you walked into the lab and said, ‘I’ve got to wait there for two hours to get my blood drawn?’ Multiple times I get patients that will come up to my clinic one hour late, and they say, ‘Unfortunately, one of the phlebotomists got sick and was not down there, so we had this big line waiting.’
If we have a machine like the Veebot, that is basically using ultrasound as well as infrared technology to draw blood, it’s very, very accurate, and can be done very quickly. I don’t have to worry about the robot getting sick. It could break down, but they’re very durable and things are working out well.
So some of these repetitive tasks will be done robotically. It’ll be more efficient and effective, and it takes out some of the human problems.
Has it been proven that room for error is less with robots than human error?
For some things. I mentioned how accurate they are with reading pathology slides once you teach the machine. It’s the same with CT scans and MRIs. I do believe that there’s no question the consistency is there.
I’m not saying that the human side needs to be completely taken out of it, but I can tell you this, if you look at the automotive industry, you have robots working on those assembly lines; the vast majority of the time they’re doing the heavy lifting. I think we’ll see that more often in medical care. We’ve been [reluctant to embrace] it because there seems to be the need for that human touch, not for some of these repetitive tasks.
Having said that, the human touch, will we see that gone?
I think you’ll see that gone in some places, but I don’t think you’ll see it gone when talking about counselling the patient. When it comes down to sitting with the patient when they’ve been diagnosed with a cancer or something, that needs that human touch, that empathy, that caring – the empathy that you see and feel when you talk to a good doctor.
I’m not talking about, ‘This person has great hands. This person’s smart as a whip.’ That doesn’t make a good doctor. You have to really be able to engage your patients, to look them in the eye and give them that empathy, and make them feel like they’re the only thing in the world at that time. They have to have that trust.
There is to me, a verbal contract. They trust you, you’re their doctor, that’s written in stone. A robot will not do that. I don’t think we’re there yet, and we won’t be there for a long time. I don’t think that humans will allow that.
Can you explain a little more the impact of initial deep learning and big data?
The point about big data and deep machine learning, for instance, is in my clinic I spend an inordinate amount of time going online with calculators. For instance, we have what we call a Kattan nomogram. I can plug in the patient’s PSA, their clinical stage and it spits out the patient’s risk for having cancer.
We’re much further along than that. Algorithms of the future will include, for instance, a number of genes, proteomics and we’ll be able to extract from the electronic medical record their family history, their social history, what are they doing that may increase their risk for cancer. What are the genes that put them at increased risk?
It’s going to take a computer very little time. It would take me years probably, to really compute, these are the people at greatest risk, because of this genetic abnormality, these proteomics, their past medical history, their social history.
That’s where the big data and more data points can be crunched, and using deep machine learning can spit out information to both a physician and the patient.
So I do think that big data is going to change the way we do business, and I think it’s going to change the information the patient’s able to get very quickly.
Image credit: Kittipong Jirasukhanon & Pop Nukoonrat via 123RF