I had a medical mystery. One day I experienced a lump that would swell and later recede. I Googled and Googled and Googled some more. I did numerous Internet searches to get an idea of anything that might cause lymph nodes to swell. It is generally an infection, and rarely, even cancer. I got a new patient appointment with my wife’s doctor, as mine had retired. It took several weeks to get in. The condition was gone when the doctor tried to examine it. She ordered an ultrasound which I got a week later, but the lump had receded when they tried to scan it. A couple weeks later I went to a specialist to see if he could shed some light on the condition. Again, there was nothing for him to see. Both doctors had theories but nothing definitive. Along the way I also received blood tests, CBC, urinalysis, and a test for markers of multiple cancers. I was motivated to have a colonoscopy, an Upper GI, and more tests just to see if something could be the cause of lymph nodes swelling.
Why did it take three months of searching, two doctors, an ultrasound scan, and multiple tests of my blood to diagnose my medical mystery? Mostly because my doctors only had about 15 minutes each to examine me and intermittent failures are the hardest problems to diagnose. In addition, I was not describing my condition accurately enough for AI. Once I changed my queries the most likely diagnosis popped up on my screen. My doctor’s theories were correct.
The Wall Street Journal reports that “Patients, doctors and nurses are turning to the technology for help identifying rare and undiagnosed diseases,” saying:
Patients and families are turning to AI to help pinpoint conditions. AI can be especially adept at flagging potential rare and hard-to-diagnose diseases, which may otherwise go undetected for years because doctors don’t often see them.
When it comes to disorders most doctors rarely encounter, obvious symptoms are often missed because they are just not in the forefront of your doctor’s knowledge base. Fidji Simo, a tech executive, talked about her time trying to find out what was causing her symptoms that later turned out to be postural orthostatic tachycardia syndrome:
“I was a textbook case, and doctors missed it because they were just so focused on the fact that I had had a rough pregnancy,” said Simo, who first started having symptoms after giving birth to her daughter in 2015.
Even with AI Chatbots diagnosis can still take time. Yet, a study published in JAMA found chatbots’ accuracy surpassed doctors.
[R]esearchers took 90 complex rare-disease cases that had already been solved and tested AI chatbots’ abilities to suggest the right diagnosis. The two chatbots got the diagnoses right in 13% and 10% of cases, respectively, compared with 5.6% of cases that doctors had diagnosed through a review of patients’ medical records.
AI is great at identifying patterns, but there are still limitations. AI can translate medical jargon, compile research, and generate leads, but doctors still need to pursue the leads to arrive at a diagnosis. One advantage AI has is speed. It can scour the medical literature in seconds that may otherwise require a dozen doctor visits spread across months or years. AI can analyze patient medical records and look for clues far faster than a physician scanning a computer screen wading through years of patient visits. AI can be trained on medical literature to a degree that doctors would never be able to digest. AI can also access and retain the information that was long forgotten by a human. AI is an invaluable tool, but it is not a doctor.