Doctors are worried, and rightly so. Recent rapid strides in artificial intelligence threaten to remove the need for doctors to a significant extent. There is an oft-discussed scenario that in the next two to three years, well-trained nurses or other mid-level providers, assisted by artificial intelligence, will be adequate to provide most medical care. So, are we going to see widespread unemployment among medical professionals?
The answer is complex. A recent well-publicized study, also highlighted in The New York Times and published in the Journal of the American Medical Association (JAMA), demonstrated that artificial intelligence using large language models could get complex diagnoses correct more than 90 percent of the time, whereas physicians were only correct about 70 percent of the time. How can physicians embrace the technology while knowing that it may lead to the demise of their profession?
A recent case of a 15-month-old child from a small town in India illustrates this dilemma. For the last two weeks, the child had frequent falls and had stopped walking altogether over the last 10 days. The child’s mother became distraught when she noticed that the child’s body was trembling and the eyes had rapid jerking movements. The pediatricians and neurologists in the town were aware that there was something seriously wrong. They had access to a proprietary AI rule engine integrated with a large language model. They entered the symptoms, and within minutes, they had a diagnosis: “dancing eye syndrome” or opsoclonus-myoclonus (OMS) syndrome. This rare condition would require an urgent MRI, lumbar puncture, and treatment with high-dose steroids and perhaps immunosuppressants. The child was seen by an appropriate specialist in a larger city, who agreed with the diagnosis. The child was promptly treated and is hopefully on the road to recovery.
The potential to save lives, make prompt diagnoses in rare and life-threatening conditions, and provide more accurate treatments based on the latest guidelines is, after all, what any caring and empathetic doctor would want for their patients. On the other hand, physicians have put in years of hard work, sacrifice, and sleepless nights to see their jobs threatened by computer algorithms. From a macro perspective, patients in underserved areas can get expert opinions and treatment advice in minutes. With reduced need for health care personnel, costs will no doubt decrease, easing the burden on countries struggling with aging populations.
I was a skeptic. I have been in medicine all my adult life, having started medical school in India at the age of 17. I then went on to train at UCLA in Los Angeles and worked in Singapore and Boston. I am strictly old school; I believe in the power of the thorough history and physical exam as taught to me by my professors. I believe in empathetic listening, the power of the detailed physical exam, and being present for patients. However, over the last year, I have served as medical advisor to an artificial intelligence health company with a proprietary AI rule engine that preprocesses medical data and feeds it to the large language model (LLM), ensuring accuracy and consistency of results. Having worked with this, I cannot be more amazed at the breadth and scope of the systems and what they can achieve for our patients. Often, patients must go from specialist to specialist to get an opinion. However, the AI systems have a depth of knowledge that few physicians can hope to achieve.
I recently saw a 25-year-old woman who had been having pain in her collarbone after pushing a door. She was seen by an orthopedic doctor who prescribed painkillers and physiotherapy. She did not improve, the pain worsened, and she noticed some swelling around her bones. In addition, she started having rashes on her body. She had always complained of back pain. Her investigations, including X-rays, inflammatory markers, and rheumatoid tests, all came back normal. She saw five doctors with no diagnosis or relief. Her symptoms were fed into the AI rule engine, and the diagnosis suggested considering a rare autoimmune entity called SAPHO syndrome. An MRI confirmed the diagnosis, and the patient is being treated for it.
So, we have seen that artificial intelligence can make better diagnoses and guide treatment as per established guidelines, even compared to seasoned medical professionals. But what about empathy and the human connection? We talk about this so much, but we, as physicians, must recognize that our patients are frustrated. They leave their jobs in the middle of the day, commute for an average of one hour, and wait in doctors’ offices for what is often a 15-minute consult. Doctors these days are distracted by the computer, inputting data, ordering tests, and spending less and less face time with the patient. In such a world, could a caring nurse assisted by AI diagnosis provide more empathetic and timely care?
Doctors are angry with me for saying this. However, we have seen that AI can make timely and accurate diagnoses even in rare conditions, serve underserved populations, and help reduce health care costs. We scoffed at the Google diagnoses and advice patients brought to us because they used to be vague and nonspecific, but we are threatened by AI diagnoses because we see their potential and disruptive power. Instead of being threatened or rejecting it, we perhaps should harness its power to provide better care.
One of my patients was admitted to the hospital for shortness of breath. She underwent a rapid battery of tests, including an EKG, echocardiogram, and pulmonary angiogram, all of which came back normal. Just when she seemed to be stable, she experienced sudden onset of excruciating back pain lasting a short while, became pulseless, and sadly died. The family, via ChatGPT, and the cardiologist simultaneously diagnosed a dissection of the walls of the aorta.
Whatever happens, we know we are on the cusp of the biggest changes in medicine since the advent of penicillin or the discovery of the human genome. Will we embrace technology and use it to improve medicine, or will the white coat be relegated to the dustbin forever? Will the medical fraternity resist or accept and adapt to change? The next couple of years will give us our answers.
Humeira Badsha is a rheumatologist.
