Automatic annotation of the CORD-19 dataset with PICO categories. The corpus was automatically labeled with an LSTM-CRF model trained on human-annotated PubMed abstracts from https://github.com/bepnye/EBM-NLP. Currently, titles and abstracts only are annotated using Population, Intervention and Outcome labels, as well as more fine-grained labels such as Age, Drug, Mortality and others.