…I would be surprised, or no I wouldn’t. The oncologists said “not surprised” about 13% of the time. And as you can see, the patients in that “not surprised” group have substantially higher mortality than those in the surprised group.
F. Perry Wilson, MD MSCE
The figure below needs additional annotations to make it explicit the number and plot mean. In particular being explicit if the plot is predication vs actual.
The table below could do with a caption stating (if correct) surving patients ie at day 100, xxx and yyy survived from the respective groups.
Both plots show that both humans and model aren't very good in the absolute sense of predicting death. As the best case scenario was only getting 60% of the deaths correctly for 90 days (if that was the intention).