Giving AI penalties results in better diagnoses (IMAGE)
Caption
Telling sick people that they’re healthy, can happen when a human doctor sees a patient. It also happens when Artificial Intelligence (AI) learns to diagnose disease. But giving a big penalty to an algorithm for false negatives results in much better precision, UJ researchers find. The research appears in Informatics in Medicine Unlocked, at https://doi.org/10.1016/j.imu.2021.100690
Credit
Graphic design by Therese van Wyk, University of Johannesburg. Based on Pixabay images.
Usage Restrictions
No restrictions.
License
CC0