878 followers
Conversational agents used in clinical apps could compromise patient safety if errors occur. This work uses Bayesian Long Short-Term Memory Networks to calculate model uncertainty for intent classification, with a specific emphasis on symptom checker CAs.