Saturday 21 September 2019

Patricia Casey: Will doctors be obsolete in the future, to be replaced by AI machines and robots?

Mind Matters

Stock picture
Stock picture

Patricia Casey

Just a few weeks ago the first robot news anchor was introduced to the Chinese. He (that's how it was referred to) explained that he would work tirelessly to deliver accurate and up-to-date news. Apparently humans will still be used during daylight hours, but the antisocial time slots will be covered by this robot.

Already in medicine robots are carrying out some procedures -although these are guided by human surgeons - studies are showing that diagnosis of skin cancer can be more accurately done using artificial intelligence than traditional methods.

The influential British Medical Journal recently conducted a timely written debate on whether artificial intelligence (AI) could replace doctors.

On the yes side was Prof Jorg Goldhahn from the Department of Health Science and Technology at the University of Zurich, while arguing against this proposition were, Vanessa Rampton Weiss Fellow at McGill University and Giatgen A Spinas Emeritus Professor from the University of Zurich.

Goldhahn pointed out that AI mimics human intelligence by learning, reasoning and self-correction. In certain specialties, such as radiology, dermatology and intensive care AI is better than humans at making diagnoses. It can also make prognostic predictions and even perform certain surgical procedures. In China a robot passed the national medical examination by 96 points in excess of the minimum requirement.

He is optimistic that the barriers to making machines better than doctors at diagnosis can be overcome since the problem is a technical one rather than a fundamental one because of the ability to endlessly process large volumes of data. Moreover, he argues machines are not influenced in clinical decision-making by outside influences such as, cultural preferences or different schools of thought. Rather they are based on symptoms and the combination in which they occur.

The possibility of inputting health information into a diagnostic programme is increasing with the use of personal health monitoring devices, electronic medical records, and even information on their social media platforms to give as complete a picture of their health over time as possible. He also believes that the programming errors of the designers can be overcome by regular testing and independent monitoring.

He argues that in the future doctors will be largely obsolete and will merely be assistants to the AI machines and robots.

The counter argument put forward by Rampton and Spinas is that AI will aid doctors, but can never replace them. They argue that doctors are able to deal with the patient as a whole person and also taking account of the social relationships that may influence their symptoms and their post-treatment care.

Take asthma, for instance. AI could indeed translate the symptoms into the correct diagnosis. But these are just one aspect of the illness. The other is the impact that it has on the person's life and on their family - an element that if not attended to can influence the outcome. It is difficult to foresee how AI could identify this through any algorithm. Symptoms are measurable and this is the strength of AI, but of particular relevance to psychiatry is the additional elements of nuance and non-verbal communication that enter into the diagnostic process.

For example, the suspicion that a person directs to the doctor during the consultation may be understandable or it may indicate an underlying paranoid disorder. The feeling of the undertone is nothing more than that, but nevertheless the experienced doctor will note and bear it in mind when arriving at a list of possible diagnoses.

There are some who have "somatoform" disorders. These are conditions with multiple physical symptoms, but with no physical or biological explanation. Entering data on these into an AI algorithm would result in a diagnosis of cancer or some other serious condition when the explanation may lie elsewhere. These are the elements in clinical practice that have led to the recognition that medicine is an art.

And what about the doctor-patient relationship, the care and empathy that the sick person needs? Of course some people with minor ailments may not necessarily require empathy, but a treatment that works, for example, an ear infection. Yet throughout the history of medicine the doctor-patient relationship has been acknowledged as being crucial to the diagnostic and treatment process. Indeed in studies of psychological interventions the relationship to the therapist is one of the key elements influencing outcome. The robotic voice may offer sympathy as a matter of form, but cannot answer questions like, 'why me'?

A Brave New World future could be on our doorstep should it come to pass that robots and machines will replace doctors. Yet an online poll among the BMJ readership overwhelmingly voted against the motion (80pc).

Health & Living

Editors Choice

Also in Life