• Guy Tsafnat

EBM Bonus Episode

This is the final episode in our series on EBM. In this series we highlighted some of the challenges facing the advancement and use of EBM. This time we are looking at four areas where technology and AI can make EBM better.

Area 1: interoperability

While traditionally in IT circles, interoperability means compliance with standards, protocols and APIs, healthcare data poses new challenges that require a bit more: common semantics, intelligently understanding of unstructured data (e.g. free text and images) for adding context to data, and lots and lots of meta-data.

Area 2: research automation

Much of the work of researchers is manual, labour intensive, cognitive, and error prone. A good starting point is for systems that mimic researchers but many improvements can be made from there. Updates of research results need to be routine and effortless.

Area 3: recycling

Research waste is here to stay. Many factors can make research waste. Many of them are hard or impossible to predict. Sometimes it doesn't take much to turn good research to waste. Research should be flexible enough so that peer review can lead to improvements in the study. Research protocols corrected and repeated without the experiment needing to start from scratch.

Area 4: communities

Communities of research need to adjust. Rather than being conducted in labs or departments, research needs to take place among labs and departments. Across the world and across disciplines. Compared to other disciplines (e.g. particle physics), medical research is lagging in this area.

#ebm #researchautomation #evidence #valuebasedcare #future

The Automation of Systematic Reviews