Members of the CAIRIBU Machine Learning interest group were invited to participate in a panel discussion at the 2024 AUGS Pelvic Floor Disorders Week in Washington DC in October 2024. The Panel discussed Practical and Ethical Considerations for AI in Urogynecology. Panelists included Giulia Ippolito, MD*; Jim Hokanson, PhD*, Whitney Hendrickson, MD*, David Sheyn, MD* and Susannah Rose, PhD (* denotes CAIRIBU members).
The clinical case-based discussion provided a general overview of ethical and practical considerations when evaluating prediction models (concordance, calibration, validation), Generative and Ambient AI applications, AI augmentation of diagnostic findings.
The takehome points were:
- Consider AI applications as clinical interventions, before use the AI tool must be validated and calibrated in the intended patient population.
- Generative AI is built on data that we have collected, are we collecting the correct data to create models to help improve patient care?
- AI is being used in our workflows (prediction calculators), consider how AI impacts and may bias your clinical decision-making
- Consider:
-
- 1) How will AI be integrated into workflows?
- 2) How will the AI impact patients – negatively and positively?
- 3) When and how will you discuss the use of AI with patients?