While doctors are certainly highly-trained and skilled in a variety of medical disciplines, multiple experiments and studies have shown that doctors are not as good decision-makers as they think they are - regardless of the amount of rigorous training they go through, they are just as prone as any other human is to flawed cognitive processes.
For example, doctors often fall prey to the availability bias, where they place particular weight on diagnoses and treatments that come to mind more easily, which can be misleading. While such biases can be actively avoided, there are certain things that even senior doctors find difficulty to keep ahead of, such as reading and filtering through vast amounts of growing healthcare data.
Rise of the machines could overcome these weaknessesArtificial intelligence (AI) has managed to bypass the humble path to wisdom that every physician has had to walk. Feeding off millions of medical evidence reports, patient records, clinical trials, and medical journals, these software easily outperform even the most experienced of doctors.
There have been computerised clinical decision-support systems that could pinpoint causes of acute abdominal pain with an accuracy comparable to that of a senior clinician, and others that could pick up cases of heart attacks that were missed by experienced cardiologists.
One particular AI that garnered attention after winning the US-television game show Jeopardy in 2011 – the Watson AI system by IBM is designed to gain more knowledge with time, learn from its success and failures, and refine its analysis based on what it is learning. It can be ‘trained’ with data from a certain field and very quickly supersedes clinicians who have spent decades training in that area.
Earlier this year, Watson famously managed to diagnose a Japanese woman’s leukeamia in just minutes, when doctors had been stumped had been unable to successfully treat her prior to that.
Still a need for robots to be improvedBut beyond these few success stories and the myriad of different software aids on the market, there are reasons why these ‘clinical wonders’ are not yet to be fully integrated into medical practice.
Performance of these aids varies widely across different conditions and usage patterns. Furthermore, there has not been a ‘model’ software to be a point of reference when measuring performance. Technicalities aside, AIs also raise concerns of patient confidentiality and privacy of medical records. AIs are like screening tests – detecting more diagnoses could bring more harm than good to the patient - especially if the system picks up indolent health problems and trigger unnecessary psychological stress.
Computer scientist and physician Tobias Mueller of the University Clinic Marburg in Germany is involved in the Rhön-Klinikum pilot study that seeks to utilise Watson for diagnosing over 7,000 rare diseases. He says that sometimes the information sources fed into Watson contradict each other.
Put another way, these sophisticated computerised diagnosis aids can still come up against some of the exact problems humans do when sharing and comparing information.
But the presence of doctors often marks beginning of the cure“While it’s true that computer recall is always going to be better than that of even the best doctor, what computers can’t do is communicate with people. People describe symptoms in very different ways depending on their personalities,” said Clare Aitchison, a medical practitioner from Norwich.
The fact remains that currently, computerised medical aids do not quite measure up to their human counterparts, according to recent studies. Additionally, a research team found broad performance variations across different diseases, as well as different usage patterns among doctors who utilised such software.
Any practicing doctor can testify that important skills in the profession include excelling at social relations and emotional intelligence. No two patients are ever the same, even if they suffer from the same condition.
Medicine continues to require the human touch - it takes the five senses to sift through different descriptions of symptoms and tailor medical practices to suit each unique patient. MIMS
The battle of doctors vs. machines in clinical diagnoses
AI Watson and other nanorobots to help ‘detective doctors’ solve medical mysteries
Where are the fault lines? A look at the causes and prevention of diagnosis errors
AI known as CRyPTIC can diagnose drug-resistant tuberculosis in minutes