Algorithmic Bias

Generalizing an Antibiotic Recommendation Algorithm for Treatment of Urinary Tract Infections to an Urban Academic Medical Center
An AI-based system for UTI antibiotic selection generalized fairly well to data from NYU Langone Health. However, utility in practice unclear due to a concerning trend of disparities against non-white patients. The tradeoff between inappropriate antibiotic treatment (IAT; how often the prescribed antiobiotic is ineffective) and broad-spectum usage (BSU; how often a stronger, more broad antibiotic is prescribed) is improved by the AI but white patients benefit significantly whereas non-white groups continue to receive ineffective and aggressive treatment.
Garrett Yoon, Richard S. Matulewicz, Vincent J. Major
Cite DOI Journal
Probing Automated Treatment of Urinary Tract Infections for Bias: A Case-Study Where Machine Learning Perpetuates Structural Differences and Racial Disparities
Urinary tract infections (UTI) are a common indication for antibiotic treatment worldwide. Fluoroquinolone antibiotics are widely …
Garrett Yoon, Vincent J. Major
PDF Cite Poster DOI Proceedings