The medical field has identified around 10,000 "rare diseases" that are estimated to affect fewer than 200,000 Americans each. On net, 10% of the population is estimated to be afflicted by a rare disease, with rampant underdiagnosis. It's less "misdiagnosis" in that the picture is being identified clearly as something other than what it is, but that few doctors have the entire picture, recognize it as one unified syndrome, and recall the title of that syndrome. The mind also loses most information that goes unused, so a disorder that a doctor only encounters once per every thousand years of practice would need a great deal of luck to be recognized, especially by someone who isn't just out of med school. Even something seen only every hundred years or every ten years of practice is usually missed. There are lots of zebras out there that we're calling horses.
https://www.ncbi.nlm.nih.gov/books/NBK56184/
> To have a rare disease is often to have a condition that goes undiagnosed for years while concerned physicians who have never seen the condition before may offer one diagnosis and then search for another when new or advancing symptoms belie the original diagnosis. Once accurately diagnosed, patients with rare conditions may be treated by physicians who have little evidence or guidance to help them—physicians who may experience the frustration imagined by the patient quoted above. Particularly when a condition is extremely rare, patients and families frequently have to travel long distances to consult with the few experts who have experience in treating and studying their rare diseases; patients and their families may even relocate to make access easier. Although the features of specific rare diseases can differ in myriad ways, the effects on life and functioning are often similar and are emotionally and financially devastating for the affected individuals and their families. Patients and family members may feel isolated and alone as they face the challenges of finding helpful information, learning a new medical language, and generally charting their way in a daunting new world.
> As described in Chapter 1, some rare conditions are extremely rare, found in only a few or a few dozen people. Others occur in hundreds, thousands, or as many as 200,000 people in the United States. Many are genetic in origin or have a genetic component. Others arise from exposure to infections or toxins, from faulty immune responses, or occasionally from adverse responses to therapeutic interventions for other conditions. For many rare conditions, the causes are frustratingly elusive.
> Although people may think of a rare disease as something that happens to someone else, rare diseases can afflict anyone, at any age. They can be acute or chronic. Many are debilitating and present an ongoing risk of death. Some are inevitably fatal given current medical options. Approved therapies are available to treat several hundred of these conditions, but most currently have no therapy that cures or modifies the disease itself.
One of the things that various AI expert systems would be great at compared to human doctors, is recognizing the possibility of rare diseases based on abundant data (or full body scans, for that matter) and proactively testing hypotheses. But... well... the only people with access to that data, the insurance companies, are strongly incentivized not to.