39 Making MetaDataCount Computer Vision News Computer Vision News Electronic Healthcare Records (EHR), particularly its impact on intersectional subgroups. The multimodal nature of EHR data—integrating text, time series, tabular information, events, and images—adds complexity to bias mitigation, as the impact on minority groups varies across modalities. The speakers first discuss how to learn unified representations from heterogeneous data, then highlight the limitations of conventional bias mitigation strategies in such multimodal settings. Next, they examine the role of bagging strategies in improving fairness using MIMIC-Eye and MIMIC-IV ED datasets. If you want to learn more about their strategies for fostering more inclusive and equitable multimodal AI systems in healthcare, check out the video above. We are planning to hold our next webinar in May 2025; sign up for the series’ newsletter if you want to stay updated!
RkJQdWJsaXNoZXIy NTc3NzU=