MT-Clinical BERT: Scaling Clinical Information Extraction with Multitask Learning
This addresses the bottleneck of managing multiple systems for clinical information extraction, which is important for healthcare applications, but it is incremental as it builds on existing BERT and multitask learning approaches.
The paper tackled the problem of limited training data and disjoint development in clinical information extraction by developing a single multitask learning model that performs eight clinical tasks, achieving competitive performance with state-of-the-art task-specific systems and providing computational benefits at inference.
Clinical notes contain an abundance of important but not-readily accessible information about patients. Systems to automatically extract this information rely on large amounts of training data for which their exists limited resources to create. Furthermore, they are developed dis-jointly; meaning that no information can be shared amongst task-specific systems. This bottle-neck unnecessarily complicates practical application, reduces the performance capabilities of each individual solution and associates the engineering debt of managing multiple information extraction systems. We address these challenges by developing Multitask-Clinical BERT: a single deep learning model that simultaneously performs eight clinical tasks spanning entity extraction, PHI identification, language entailment and similarity by sharing representations amongst tasks. We find our single system performs competitively with all state-the-art task-specific systems while also benefiting from massive computational benefits at inference.