Systematic reviews and systematic maps are crucial tools for analysing the breadth of evidence in the livestock and animal health space. However, traditional methods are labour-intensive, requiring researchers to manually screen thousands of papers and extract data – a process that can take months or even years. 

Our Solution: an end-to-end pipeline 

Developed in collaboration with EDINA, the University of Edinburgh’s centre for data and digital expertise, LitXpress is an automated tool designed to produce rapid reviews in weeks rather than years. It mirrors the core data pipeline of SEBI-L, applying Machine Learning to unstructured literature:

1. Find: LitXpress currently automates searches across four bibliographic databases, significantly reducing the time spent querying and downloading results. Further testing is needed, using more varied queries, to help determine whether additional databases could be integrated to improve comprehensiveness.
2. Classify: Using machine learning, the tool classifies articles by relevance while still allowing users to record their own classifications and compare them with the tool. Early testing shows promising precision, allowing researchers to focus on validation rather than discovery. Further improvements will enhance time saving by reducing reliance on manual screening. This will be done primarily through ongoing work to refine the classifier and increase confidence in assessments of article relevance.
3. Extract & Consolidate: The current system extracts key data points and outputs a simple structured data file ready for analysis. Future iterations will use machine learning to improve processing of extracted data after human classification. This year, we will begin work on integrating LitXpress outputs with CAMARADES tools, to streamline data extract cleaning and enable more complex, hierarchical data outputs (e.g., linking values to specific species).

Testing, learning, improving 

LitXpress is currently a minimum viable product focused on animal disease prevalence. Our vision is to build an end-to-end rapid review tool which can be trusted to generate fit-for-purpose data within a short amount of time. 

We will take advantage of machine learning advancements at the University of Edinburgh and draw on substantial developments in the automation of systematic reviews tools from other sectors notably human health to help LitXpress realise its potential.

As we work on improving data generation and usability and expand into broader areas of livestock health and production, we remain committed to transparency, sharing our progress from prototype to mature tool.

More information

• Learn more about SOLES and SyRF (our collaboration with CAMARADES)
• Visit the EDINA website