Artifact Under Review
FMEA and AI bias
Team Number for Team Doing the Review
Team 27
Description of Issue
I think it might be worthwhile to discuss how AI bias might affect the project. AI bias can lead to inaccurate diagnoses for certain patient groups, which is a significant safety and reliability concern. I would suggest adding AI bias as a specific hazard in the FMEA. This could include failure modes like biased training data or lack of diversity in the dataset. For mitigation, you could recommend actions such as diversifying the training datasets, implementing fairness algorithms, and regularly evaluating the model’s performance across different demographics.
RezaJodeiri
commented
3 hours ago
Artifact Under Review FMEA and AI bias Team Number for Team Doing the Review Team 27 Description of Issue I think it might be worthwhile to discuss how AI bias might affect the project. AI bias can lead to inaccurate diagnoses for certain patient groups, which is a significant safety and reliability concern. I would suggest adding AI bias as a specific hazard in the FMEA. This could include failure modes like biased training data or lack of diversity in the dataset. For mitigation, you could recommend actions such as diversifying the training datasets, implementing fairness algorithms, and regularly evaluating the model’s performance across different demographics.