Automated extraction of data for medical systematic reviews from variable-quality PDF documents
Lead Research Organisation:
University of Nottingham
Department Name: School of Computer Science
Abstract
Systematic reviews are a form of medical literature review designed to survey all extant research on a particular topic, in order to make recommendations to practitioners and policy makers as to what treatments should be offered based on the best-available evidence. Such reviews are very time-consuming to produce, and many of the tasks involved in writing systematic reviews are suitable for either automation or machine support. One such task is the extraction of data from reports of medical trials, typically held in PDF form, which will be used to inform the review's conclusions. This project is intended to develop new methods of providing machine support in this process, which will speed the review production process and contribute towards the goal of full automation of review production.
Organisations
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/N50970X/1 | 01/10/2016 | 30/09/2021 | |||
| 2049919 | Studentship | EP/N50970X/1 | 01/10/2017 | 30/12/2020 | Ian Knight |
| Title | Automated data assistant for systematic review data extraction |
| Description | A web-based tool for presenting machine learning-generated predictions to a user performing data extraction for systematic review. The tool serves two purposes: it functions as a prototype of a fully-fledged system we hope to deploy, and also as a tool to investigate user perceptions of different methods of interacting with machine learning predictions and automation. |
| Type Of Technology | Webtool/Application |
| Year Produced | 2020 |
| Impact | The study this software relates to is currently being finalised, so there are no outputs as yet. |
| URL | http://cszg.nottsknight.uk/data-assistant |