TI Methods Speaker Series: New models for evidence synthesis using machine learning/artificial intelligence

Evidence synthesis is a key method for ensuring clinical practice is up-to-date, efficient, and effective; but conventionally done systematic reviews are laborious, time-consuming, and quickly become out of date. Developments in artificial intelligence (AI) systems mean that much of the manual work of evidence synthesis could be automated. These technologies move beyond speeding up a conventional systematic review, and could enable new models of evidence synthesis where research databases are continually monitored by AI systems. These developments raise the prospect of fully automatic, living evidence synthesis, and new guideline models where recommendations can update rapidly on publication of new research evidence. This talk will discuss these developments, illustrated by the RobotReviewer and Trialstreamer systems, and describe what could be possible now, and in the future.

About the speaker: Iain Marshall is a senior lecturer at King’s College London, and south London GP, whose research focuses on using computers to speed up the process of getting evidence into practice. Iain co-leads the RobotReviewer project, which is an NIH-funded study investigating the use of natural language processing and machine learning for evidence synthesis.