Multi-label classification for biomedical literature: an overview of the BioCreative VII LitCovid Track for COVID-19 literature topic annotations.

The coronavirus disease 2019 (COVID-19) pandemic has been severely impacting global society since December 2019. The related findings such as vaccine and drug development have been reported in biomedical literature-at a rate of about 10 000 articles on COVID-19 per month. Such rapid growth significantly challenges manual curation and interpretation. For instance, LitCovid is a literature database of COVID-19-related articles in PubMed, which has accumulated more than 200 000 articles with millions of accesses each month by users worldwide. One primary curation task is to assign up to eight topics (e.g. Diagnosis and Treatment) to the articles in LitCovid. The annotated topics have been widely used for navigating the COVID literature, rapidly locating articles of interest and other downstream studies. However, annotating the topics has been the bottleneck of manual curation. Despite the continuing advances in biomedical text-mining methods, few have been dedicated to topic annotations in COVID-19 literature. To close the gap, we organized the BioCreative LitCovid track to call for a community effort to tackle automated topic annotation for COVID-19 literature. The BioCreative LitCovid dataset-consisting of over 30 000 articles with manually reviewed topics-was created for training and testing. It is one of the largest multi-label classification datasets in biomedical scientific literature. Nineteen teams worldwide participated and made 80 submissions in total. Most teams used hybrid systems based on transformers. The highest performing submissions achieved 0.8875, 0.9181 and 0.9394 for macro-F1-score, micro-F1-score and instance-based F1-score, respectively. Notably, these scores are substantially higher (e.g. 12%, higher for macro F1-score) than the corresponding scores of the state-of-art multi-label classification method. The level of participation and results demonstrate a successful track and help close the gap between dataset curation and method development. The dataset is publicly available via https://ftp.ncbi.nlm.nih.gov/pub/lu/LitCovid/biocreative/ for benchmarking and further development. Database URL https://ftp.ncbi.nlm.nih.gov/pub/lu/LitCovid/biocreative/.

Text and Network-Mining for COVID-19 Intervention Studies (preprint)

Background: The COVID-19 pandemic has led to a massive and collective pursuit by the research community to find effective diagnostics, drugs and vaccines The large and growing body of literature present in MEDLINE and other online resources including various self-archive sites are invaluable for these efforts. MEDLINE has more than 30 million abstracts and an additional corpus related to COVID-19, SARS and MERS has more than 40,000 literature articles, and these numbers are growing. Automated extraction of useful information from literature and automated generation of novel insights is crucial for accelerated discovery of drug/vaccine targets and re-purposing drug candidates.Methods: We applied text-mining on MEDLINE abstracts and the CORD-19 corpus to extract a rich set of pair-wise correlations between various biomedical entities. We built a comprehensive pair-wise entity association network involving 15 different entity types using both text-mined associations as well as novel associations obtained using link prediction. The resulting network, which we call CoNetz, also contains a specialized COVID-19 subnetwork that provides a network view of COVID-19 related literature. Additionally, we developed a set of network exploration utilities and user-friendly network visualization utilities using NetworkX and PyVis.Results: CoNetz consisted of pair-wise associations involving 174,000 entities covering 15 different entity types. The specialized COVID-19 subnetwork consisted of 7.8 million pair-wise associations involving 43,000 entities. The network captured several of the well-known COVID-19 drug re-purposing candidates and also predicted novel candidates including ingavirin, laninamivir, nevirapine, paritaprevir, pranlukast and peficitinib.Conclusions: Our automated text and network-mining approach builds an up-to-date and comprehensive knowledge network from literature for COVID-19 studies. The wide range of entity types captured in CoNetz provides a rich neighborhood context around the relations of interest. The approach avoids multiple drawbacks associated with manual curation including cost and effort involved, lack of up-to-date information and limited coverage. Amongst the novel repurposing drugs predicted, laninamivir and paritaprevir are possible COVID-19 anti-viral drugs while pranlukast was postulated to be a candidate for managing severe respiratory symptoms in COVID-19 patients. CoNetz is available for download and use from https://web.rniapps.net/tcn/tcn.tar.gz