Please enable JavaScript.
Coggle requires JavaScript to display documents.
Bias Mitigation, Introduction, AI for Medical Use, Medical biases, AI for…
-
Introduction
-
-
-
Doctors are human beings, who might hold biases even if they aren't malicious (not racist)
While a human is slow to unlearn biases, Data Driven computerized assistance can correct for biases, thus giving a potentially fair-er diagnosis or at least recommendation to the medical professional
AI for Medical Use
Bias in AI
AI isn't immune to bias - 'Garbage in, garbage out'
-
Multiple methods for bias correction are possible - at different times of pipeline (3) and by optimizing different metrics (6)
-
-
-
Medical biases
Gender bias
women are perceived to be less prone to some diseases, doctors don't seem to trust them when they complain and are slower to get prescribed medication/treatment
-
Racial bias
-
TO DO: check again about other races, I think they arent as big victims as black ppl but need source
-