Methodological Rigor in Surgical AI: Bias, Validation, and a Multimodal Risk Model in Development
Event Description
Educational objectives:
Speaker: Dr. Anisha R. Kumar, MD Clinical Assistant Professor, Surgery and Biomedical Informatics, Stony Brook University School of Medicine
Location: MART Building, Room 7M-0602 (7th Floor)
Remote Access: https://stonybrook.zoom.us/j/95617197636?pwd=KytzZ2pVRG9SZGpKZUtpNXJISjNjZz09Meeting ID: 95617197636 | Passcode: 924293
- Describe how bias enters AI systems across the development pipeline, using aesthetic facialevaluation as a case study, and identify mitigation strategies applicable at each stage.
- Apply the PROBAST+AI risk-of-bias framework to critically appraise the methodologicalquality of a published prediction model using regression or machine learning methods.
- Compare the discriminative performance of multiple modeling architectures against anexisting clinical risk calculator on a large benchmarking dataset.
- Recognize the complementary roles of clinical expertise and informatics methodology indefining clinically meaningful AI project objectives, predictors, and evaluation criteria.
Speaker: Dr. Anisha R. Kumar, MD Clinical Assistant Professor, Surgery and Biomedical Informatics, Stony Brook University School of Medicine
Location: MART Building, Room 7M-0602 (7th Floor)
Remote Access: https://stonybrook.zoom.us/j/95617197636?pwd=KytzZ2pVRG9SZGpKZUtpNXJISjNjZz09Meeting ID: 95617197636 | Passcode: 924293
