Reading this story about new story about AI in medicine reminded me of a dear friend of the family who passed away several yeas ago. D was an astrologer of considerable repute and getting an audience with him took many months of effort. He predictions were accurate and his demeanor calming. People usually went away feeling better after they saw him even if the news was not that great.
D routinely predicted when a person might experience a cardiac event and likelihood of surviving it. I know of a few people who made it like he said they might and others that did not. He was older than my grandfather and lived to be almost hundred years old. The man was an institution and had helped generations of people - his approach was very methodical. They article talks about a neural network to learn from 10 years of standard clinical patient data, 22 factors such as patients' age, weight, race and prescription drug use
In similar vein D used a large number of data points that no scientist might agree with and used established outcome patterns that were related to combinations of those data points to make predictions about cardiac events. Many would argue that astrology is a pseudo-science and I would be in their ranks had I not had the privilege for knowing D for as long as I did. There are some inherent similarities in approach. Who is to say that one set of data points is totally irrelevant even if both paths to answering the question yields the same correct answer.
Comments