Validation of a Natural Language Machine Learning Model for Safety Literature Surveillance

ConclusionCharacterizing model performance prospectively, under real-world conditions, allows us to thoroughly examine model consistency and failure modes, qualifying it for use in our surveillance processes. We also identify potential future improvements and recognize the opportunity for the community to collaborate on this shared task.
Source: Drug Safety - Category: Drugs & Pharmacology Source Type: research