ASGE Masterclass: Artificial Intelligence (Live Virtual) | June 2021-Question & Answer

Video Transcription

Video Summary

In this video, several speakers discuss the issue of bias in medical data and algorithms. Yuichi emphasizes the importance of diversity in data collection to prevent bias. Mike highlights the need to improve the infrastructure for data collection in hospitals. He also suggests using machine learning for automatic labeling and crowdsourcing labeling to improve generalizability. Taposh mentions the need for better algorithms and generating artificially the desired images. Sravanti discusses the challenges of using algorithms from different populations and recommends post-market analysis and centralized certification for algorithms. In conclusion, Yuichi notes the lack of major differences between ethnicities in colonoscopy, but emphasizes the need to address bias in other healthcare areas. No credits are mentioned.

Keywords

bias in medical data

diversity in data collection

improving data collection infrastructure

machine learning for automatic labeling

challenges of using algorithms from different populations