Snorkel AI
Case Studies
Scaling Clinical Trial Screening at MSKCC with Snorkel Flow
Overview
Scaling Clinical Trial Screening at MSKCC with Snorkel FlowSnorkel AI |
Sensors - Flow Meters Sensors - Liquid Detection Sensors | |
Cement Education | |
Product Research & Development Quality Assurance | |
Tamper Detection Virtual Training | |
Testing & Certification Training | |
Operational Impact
The document classification AI application built by the team is now used downstream to power a clinical trial screening system. This system allows MSKCC to identify HER-2 among patient records without relying on human experts to review each record. The use of Snorkel Flow has significantly reduced the time to label complex, domain-specific text documents as training data by labeling programmatically. It has also increased explainability by encoding the labeling rationale for each training data point as labeling functions that can be inspected like code. The team was able to use model-guided error analysis to identify data quality issues and iterate rapidly to improve. | |
Quantitative Benefit
Achieved an overall accuracy of 93% and an average F1 of 87% across all classes | |
Auto-labeled thousands of patient records | |
Reduced time to build a document classification from months to weeks | |