> Experienced radiologists who evaluate chest X-rays as "normal" or "abnormal" contradict themselves 20% of the time when they see the same picture on different occasions
Seems like computers/machine-learning could handle a majority of diagnoses, with a human being needed to come in and reevaluate the tricky edge cases.
http://books.google.com/books?id=TA7Q27RWlj0C&q=radiolog...
> Experienced radiologists who evaluate chest X-rays as "normal" or "abnormal" contradict themselves 20% of the time when they see the same picture on different occasions
Seems like computers/machine-learning could handle a majority of diagnoses, with a human being needed to come in and reevaluate the tricky edge cases.