Specialised pre-trained language models are becoming more frequent in Natural language Processing (NLP) since they can potentially outperform models trained on generic texts. BioBERT (Sanh et al., Distilbert, a distilled version of bert: smaller, faster, cheaper and lighter. , 2019) and BioClinicalBERT (Alsentzer et al., Publicly available clinical bert embeddings. In , pp. 72-78, 2019) are two examples of such models that have shown promise in medical NLP tasks. Many of these models are overparametrised and resource-intensive, but thanks to techniques like knowledge distillation, it is possible to create smaller versions that perform almost as well as their larger counterparts. In this work, we specifically focus on development of compact language models for processing clinical texts (i.e. progress notes, discharge summaries, etc). We developed a number of efficient lightweight clinical transformers using knowledge distillation and continual learning, with the number of parameters ranging from million to million. These models performed comparably to larger models such as BioBERT and ClinicalBioBERT and significantly outperformed other compact models trained on general or biomedical data. Our extensive evaluation was done across several standard datasets and covered a wide range of clinical text-mining tasks, including natural language inference, relation extraction, named entity recognition and sequence classification. To our knowledge, this is the first comprehensive study specifically focused on creating efficient and compact transformers for clinical NLP tasks. The models and code used in this study can be found on our Huggingface profile at https://huggingface.co/nlpie and Github page at https://github.com/nlpie-research/Lightweight-Clinical-Transformers, respectively, promoting reproducibility of our results.

We propose a new setting for question answering in which users can query the system using both natural language and direct interactions within a graphical user interface that displays multiple time series associated with an entity of interest. The user interacts with the interface in order to understand the entity's state and behavior, entailing sequences of actions and questions whose answers may depend on previous factual or navigational interactions. We describe a pipeline implementation where spoken questions are first transcribed into text which is then semantically parsed into logical forms that can be used to automatically extract the answer from the underlying database. The speech recognition module is implemented by adapting a pre-trained LSTM-based architecture to the user's speech, whereas for the semantic parsing component we introduce an LSTM-based encoder-decoder architecture that models context dependency through copying mechanisms and multiple levels of attention over inputs and previous outputs. When evaluated separately, with and without data augmentation, both models are shown to substantially outperform several strong baselines. Furthermore, the full pipeline evaluation shows only a small degradation in semantic parsing accuracy, demonstrating that the semantic parser is robust to mistakes in the speech recognition output. The new question answering paradigm proposed in this paper has the potential to improve the presentation and navigation of the large amounts of sensor data and life events that are generated in many areas of medicine.

The generation of referring expressions is a central topic in computational linguistics. Natural referring expressions - both definite references like 'the baseball cap' and pronouns like 'it' - are dependent on discourse context. We examine the practical implications of context-dependent referring expression generation for the design of . Currently, not all spoken systems have the goal of generating natural referring expressions. Many researchers believe that the context-dependency of natural referring expressions actually makes systems usable. Using the dual-task paradigm, we demonstrate that generating natural referring expressions that are dependent on discourse context reduces cognitive load. Somewhat surprisingly, we also demonstrate that practice does not improve cognitive load in systems that generate consistent (context-independent) referring expressions. We discuss practical implications for spoken systems as well as other areas of referring expression generation.