Say you have a headache and are convinced you have a brain tumor or maybe an aneurysm. Do you want a neurologist to take a history, examine you, and explain the diagnostic and treatment plan and do so in collaboration with you and with empathy? Or, do you want to want to Google your symptoms, wait for a response and get ready for a barrage of ads for headache medications?

A no-brainer you might say, but not so fast. When MRIs became widely available, the going joke was that an MRI scanner was better than a roomful of neurologists. Referring every patient for an MRI is easy, but our healthcare system would run out of money at an even faster rate.

Machine deep learning/artificial intelligence has been shown to be more accurate than a radiologist in reading an MRI, and, than a dermatologist in recognizing skin cancer. These successes are based on pattern recognition. This can’t tell you what a patient means by “dizziness” or what “quite a while” means.

Stroke telemedicine has arrived with a thud. A neurologist still needs to look at the video and a person at the camera end must tell the patient what to do. But this has great value in delivering care to people that do not have access to a stroke center. Artificial intelligence has the potential to do the same when a neurologist is not available.

A roomful of neurologists was recently asked to vote whether artificial intelligence would replace them. As you might guess, the response was 65% to 35% that it would not.

See Neurology Today, June 6, 2019, pages 6–7.

Read More