Charleston Data Science Meetup

Understanding bias, auditing and quality control are key challenges for developing machine learning methods in healthcare. Models that have been trained on biased data, have the potential to automate decisions that are unfair and inequitable. This talk discusses strategies and methods for domain and outcome bias exploration to evaluate performance equitably across subgroups and whether the training data are representative of novel data expected in the model application. Three themes of exploration addressed are, subgroup performance equitability, application and training data similarity, training data grouping variable congruence.­­ Models developed on two publicly available datasets and a novel data set from MUSC describing mammogram screening uptake are used as examples exploring and detecting bias