Managing the evidence infodemic: Automation approaches used for developing NICE COVID-19 living guidelines

Background and ObjectivesThe National Institute for Health and Care Excellence (NICE) produces evidence-based guidance and advice for health, public health and social care practitioners in England and Wales. Between March 2020 and March 2022, NICE produced 24 COVID-19 guidelines to support healthcare workers during the COVID-19 pandemic. This article outlines three automation strategies NICE utilised to facilitate faster processing of evidence on COVID-19 and describes the value of those approaches when there is an increasing volume of evidence and demand on resources. Study Design and SettingText classification using machine learning, and regular expression-based pattern matching were used to automate screening of literature search results. Relevant clinical trials were tracked by automated monitoring of clinical trial databases and Pubmed. ResultsThe strategies discussed here brought considerable efficiencies in the processing time without impacting on quality compared to equivalent manual efforts. Additionally, the paper illustrates how to incorporate automation into established processes of the evidence management pipeline. ConclusionsWe have demonstrated through testing and use in live guideline development and surveillance that these are effective and low risk approaches at managing high volumes of evidence. Highlights- To illustrate how NICE utilised automation to handle the Covid-19 infodemic-managing the infodemic of evidence surveillance is a shared global issue. - To outline automation strategies to facilitate faster processing of evidence, especially when there is an increasing volume of evidence and demand on resources. - How automation can be included in established processes without disrupting business as usual operations. - Automation can take many forms, and depending on risk appetite, can be supplemented with manual checking. - Automation can be adopted easily with the right tools and techniques.