Hospitals of the future: Abandoned and derelict. AI will remove the need for human experts and decentralise healthcare, allowing for treatment to take place at home.

Doctors are valued for our ability to use a combination of logic, memorised knowledge and communication skills to perform three main functions. The first is to diagnose a medical condition, the second is to treat it. The third, and most important function is to communicate information to another human being with empathy, in a way that's tailored to their ability to understand, and in a way which makes them feel cared for. It's a huge part of the healing process. With increasing demand for healthcare, and with stressed out doctors stretched ever thinner, the tendency is for doctors to concentrate on the diagnosing and treating. The emotional, concerned, human aspect of treating disease is neglected, but this function is the only one even the most advanced artificial intelligence is unlikely to ever genuinely perform.

Medicine has long been considered an art, with "clinical acumen" the major skill of a doctor. He or she will generally spend time learning the relevant science and anatomy of the human body and develop skills through apprenticeship and experience treating many patients. But doctors are human and the workings of the human body are complex, and without reigning in our egos and making way for objective scientific evidence, we risk sliding down the slippery slope of illusory correlation. We unconsciously forget when a treatment hasn't worked and get excited when it appears it has. Over time, great "experience" is developed where our biased cerebral database overwhelmingly remembers the successful outcomes.

"The job of a doctor in the 21st century is becoming less of an art, and more of a human face to the precise, constant whirr of a behind-the-scenes scientific machine."

Doctors such as Alvan Feinstein, in the late 1960s, recognised the problems with relying on anecdotal evidence and wrote about the need for evidence based medicine, and Archie Cochrane, in the 1970s, wrote about the importance of using the randomised controlled trial to assess proposed new treatments against the available options. The Cochrane organisation now consists of over 37,000 volunteers systematically reviewing randomised controlled trials to provide higher quality evidence for and against diagnoses and treatments. Today as a doctor, a large part of what I do consists of coming up with a list of possible diagnoses based on signs and symptoms, then running tests and referring to established diagnostic criteria to come up with a diagnosis. Finally, evidence-based guidelines and protocols are referred to in deciding what treatment to start. The job of a doctor in the 21st century is becoming less of an art, and more of a human face to the precise, constant whirr of a behind-the-scenes scientific machine.

In the 21st century, technology startups and major players like Google are developing artificial intelligence methods to analyse data in a completely new way. The idea is to take massive amounts of information and use computer software to discover correlations between a huge number of variables. These are often a lot more complex than we can comprehend as computers do logic and impartiality far better than we do. Does coffee cause cancer? Does it reduce the risk? Does coffee increase the risk of cancer with a Mediterranean diet but reduce it with a western diet? Is there some sort of chemical reaction between two types of food that produces a chemical that increases the risk of heart attack? The scope of possible variables is mind blowing, and it amazes me that we're still considering one or two variables among millions of possible combinations and making conclusions based on our primitive testing methods and a comparitively limited capacity for logic.

Machines that can both keep up with, and draw conclusions from all knowledge and all data as it's being collected will one day replace humans in making clinical decisions. A new paradigm will emerge, in which machines will become advisors to the human front line of healthcare.

It is now becoming clear that we need to evolve further yet, from the randomised controlled trial, which is the best option we currently have, and harness the power of computers and the increasing amount of data available to prevent diseases happening in the first place. Right now, there is evidence to show that if you pig out on fast food and go out drinking every day you might not live as long as expected. We think eating green things will keep you healthy longer. But if we knew for sure, and were able to predict to the month or year the impending heart attack, stroke or even cancer based on our genetics, breath chemical profile, flow dynamics of the blood in our blood vessels, and variables collected by an AI algorithm which are beyond our comprehension, just imagine how the way we look at health could change.

I envision a future for health where computers noninvasively monitor millions of variables in billions of people in unfathomable ways, running a million parallel high frequency experiments, finding links and advising us with absolute precision what behaviour, diet change or minor medical intervention is required at any point to avoid disease.

  1. Dawes, Robyn M. "Experience and validity of clinical judgment: The illusory correlation." Behavioral Sciences & the Law 7.4 (1989): 457-467.
  2. Feinstein, Alvan R. "Clinical judgment." (1967).
  3. Ransohoff, David F., and Alvan R. Feinstein. "Problems of spectrum and bias in evaluating the efficacy of diagnostic tests." New England Journal of Medicine 299.17 (1978): 926-930.
  4. Cochrane, A. L. "Effectiveness and efficiency: random reflections on health services." London: England (1972).
  5. Google DeepMind: