I don't think these news are accurate. Predicting when someone will die is not possible, unless the AI predicts how likely a person is to suffer from a specific disease (ie. examine the arteries and check for the possibility of a heart attack), and then estimate how long people with a similar problem tend to live. And this is still pretty vague. There are just too many factors to be able to predict anything only through a chest scan. And 69% accuracy - that's a lot. The only way to have achieved this result is if they only checked patients who were already ill with terminal diseases.
You are viewing a single comment's thread from:
Hi elemenya - thank you for your comment.
As I pointed out the AI was fed CT imagery and then used machine learning techniques to detect any anomalies or strange patterns. AI considered 15,957 biomarkers and then used to make an estimation of their remaining life.
As you can see, AI did not match different people and estimated the life span. It actually assessed the person at hand.