Using neural networks to support high-quality evidence mapping

Thomas B Røst,Gunn E Vist,Laura Slaughter,Ashley E Muller,Øystein Nytrø

doi:10.1186/s12859-021-04396-x

Abstract

BackgroundThe Living Evidence Map Project at the Norwegian Institute of Public Health (NIPH) gives an updated overview of research results and publications. As part of NIPH’s mandate to inform evidence-based infection prevention, control and treatment, a large group of experts are continously monitoring, assessing, coding and summarising new COVID-19 publications. Screening tools, coding practice and workflow are incrementally improved, but remain largely manual.ResultsThis paper describes how deep learning methods have been employed to learn classification and coding from the steadily growing NIPH COVID-19 dashboard data, so as to aid manual classification, screening and preprocessing of the rapidly growing influx of new papers on the subject. Our main objective is to make manual screening scalable through semi-automation, while ensuring high-quality Evidence Map content.ConclusionsWe report early results on classifying publication topic and type from titles and abstracts, showing that even simple neural network architectures and text representations can yield acceptable performance.

Highlights

The Living Evidence Map Project at the Norwegian Institute of Public Health (NIPH) gives an updated overview of research results and publications
We received a set of training data from NIPH as a result of their manual coding to produce evidence maps for COVID-19, and we focused our research questions on exploring the classification of publications for automated coding
The choice of deep learning machine learning methods is motivated by recent performance advances and because they usually reduce the need for activities such as feature engineering [25]

Summary

Introduction

As part of NIPH’s mandate to inform evidence-based infection prevention, control and treatment, a large group of experts are continously monitoring, assessing, coding and summarising new COVID-19 publications. Policy makers and researchers worldwide are scrambling to keep up with the influx of potentially relevant COVID-19 studies. Whereas a traditional peer review- and journal-based publication process would take 6–12 months, research findings often find their way to readers in a matter of days or weeks. The use of preprint servers, with only cursory quality checks, is increasing. While this has had a positive impact on knowledge dissemination speed in the medical sciences, this arguably comes at a cost to quality, reliability and trustworthiness [1].

Objectives

Methods

Results

Discussion

Conclusion