AI-Assisted Annotator Using Reinforcement Learning

摘要: Machine learning in the healthcare domain is often hindered by data which are both noisy and lacking reliable ground truth labeling. Moreover, cost of cleaning annotating this significant since, unlike other domains, medical annotation requires work skilled professionals. In work, we introduced use reinforcement to mimic decision-making process annotators for events allowing automation Our agent learns annotate health monitor alarm based on annotations done an expert. We demonstrate efficacy our implementation ICU critical sets. evaluate algorithm against standard supervised machine deep methods. Compared SVM LSTM methods, method achieves high sensitivity that data; exhibits better generalization across mixed downsampling; preserves comparable model performance.