Finally got round to catching up with the Harvard Business Review article about Robots taking peoples jobs, in which they forecast the rise of Robot Doctors. There have been a lot of excitable headlines about Robots wiping out white collar jobs, and the HBR paper generated plenty of noise in the press.
I understand why using robots in healthcare seems like a really good idea. Technology has helped loads of industries become cheaper and more efficient, and making access to medical care cheaper and easier has a massive benefit globally. Right now I am sure Theresa May and Jeremy Hunt are being told by their Special Political Advisors that the NHS needs to become much more efficient rather than giving it more money. Black hole, service change, new technology, you can guess the rest.
I am sceptical about the ability of Robots to replace Doctors, particularly in developed healthcare markets like the UK and the US. Last year for example Johnson and Johnson discontinued their Sedasys automated sedation system. To illustrate why I am a sceptic here are a couple of scenarios:
Toni and Guy
I get my haircut in Toni and Guy. It’s a great haircut, and I enjoy the whole experience, the white walls, staff all in black, hair wash, free coffee. I could get it cut somewhere cheaper, or do it myself, but I don’t.
If Toni and Guy had a haircutting machine that was 99% as good as my current stylist (Catherine), 50% faster and 50% cheaper I still would get Catherine to cut my hair. Partly that is because of techno-fear, and partly because I have seen the haircutting machine in Chitty Chitty Bang-Bang. Mainly, however, it is because the person is the experience. I am not paying for an output (a haircut), I am paying for the time of a very good hair stylist, and a product that takes away that contact is much less valuable to me. A Robot which worked alongside my stylist and who could remember every haircut I had ever had would be much more useful, if I ever wanted to go back to a previous hair style (honestly who wouldn’t love a robot that could tell you every haircut you had every had?).
In real world healthcare Doctors aren’t dealing with definite and specific diagnosis all of the time, easily identified by bio-chemical markers. They are dealing with people in distress, who are often unsure what is actually happening to them. Getting a diagnosis and treatment is of course the outcome, but the process of how it is arrived at – the time the Doctor takes with them, the skill with which the Doctor teases out the crucial bit of information that allows diagnosis to be made – are just as important to the patient. Often they have a material impact on the outcome too.
I would have a very strong preference for a real human Doctor, no matter how fallible, over a robot in all but the most straightforward diagnosis. The NHS tried to create the equivalent of the robot who remembers my former hairstyles – it was the National Programme for IT, and it demonstrated perfectly how easy it is to waste huge sums of money trying to use widespread standardised technology to drive managerial definitions of healthcare process efficiency.
A service with fewer Doctors who spend less time with patients is worse in the eyes of the patient regardless of the clinical outcome. As a former NHS manager who spent loads of time trying to make healthcare processes more efficient it is easy to miss this point. In fact most human services have a similar dynamic, for the customer more efficient just means a bit worse.
Benefits Medical Assessments
As well as providing healthcare Doctors take responsibility for it. As professionals, answerable to the GMC for their actions, they are accountable for the decisions that they make and the care they provide. When something goes wrong it is more often the Doctor who is held to account, rather than the organisation.
Transferring this responsibility from the Doctor to the organisation is tricky, as the experience in the UK with providing outsourced medical assessments for disability benefits demonstrates. Typically the organisations which carried out these assessments had non-medically qualified assessors who worked to protocols developed by a small number of Doctors, who were clinically accountable for these protocols.
The decisions made through these systems were highly controversial, and a great many of them were overturned on appeal. Pretty much every Doctor I have spoken to has a horror story to tell about the decisions these processes produced. The assessors who were making the decisions weren’t evil, or callous, they were acting logically according to the system who employed them. The system contained an incentive to produce decisions radically different to the decisions that the patients own GP or Consultant would have made. So radically different in fact that I am surprised that none of the senior Doctors involved involved in these systems ended up in front of a GMC panel.
The legal, ethical, and insurance implications of taking clinical responsibility away from the individual Doctor and transferring it to an algorithm are much greater than people like HBR realise. The potential for organisational, legal, and commercial imperatives to distort clinical priorities through the use of robots, or any kind of algorithm based decision maker, are huge.
Personally I would spray paint Doctors silver and dress them in space age spandex. Patients would love it.