Article type
Year
Abstract
Background:
The Evidence Pipeline seeks to transform study identification for Cochrane and other evidence synthesis producers (see Figure). Research enters the Evidence Pipeline and goes through tailored workflows involving crowdsourcing and machine learning, working together to produce accurate, reliable metadata about studies. The Evidence Pipeline has been applied at both the review level and at the repository level. At the review level, a workflow called Screen4Me helps to reduce author screening burden through assessment of title-abstract records; at the repository level, studies relevant for Cochrane’s Central Register of Controlled Trials (CENTRAL) or the Cochrane COVID-19 Study Register (CCSR) are identified using Crowd and machine, helping to keep these important resources up to date.
Methods:
This presentation will provide a detailed look at the main component parts of the Evidence Pipeline applied at the specific review level (via S4M) and at the data repository level (CENTRAL and CCSR). It will also describe our evaluation work on each component part and on the system as a whole in comparison to traditional methods.
Results:
At the time of writing, the Evidence Pipeline has been used (via the Screen4Me workflow) in over 100 Cochrane reviews. The mean reduction in screening workload for author teams is 72% with a mean time to screening task completion of ten days per review. At the repository level, 99% of RCTs in CENTRAL have been identified by the Evidence Pipeline.
Conclusions:
The Cochrane Evidence Pipeline aims to expedite the study identification process for reviews and other evidence synthesis outputs, offering significant workload reduction for author teams within the current review production paradigm, but is also enabling a new paradigm to emerge—one that leverages economies of effort and scale to help create comprehensive, curated repos of studies for the EBHC communities who need it.
Patient, public, and/or healthcare consumer involvement:
The Evidence Pipeline uses Cochrane Crowd which is Cochrane’s citizen science platform. Anyone with an interest in health can join Cochrane Crowd. This collective effort has played a significant role in enabling the Cochrane Evidence Pipeline.
The Evidence Pipeline seeks to transform study identification for Cochrane and other evidence synthesis producers (see Figure). Research enters the Evidence Pipeline and goes through tailored workflows involving crowdsourcing and machine learning, working together to produce accurate, reliable metadata about studies. The Evidence Pipeline has been applied at both the review level and at the repository level. At the review level, a workflow called Screen4Me helps to reduce author screening burden through assessment of title-abstract records; at the repository level, studies relevant for Cochrane’s Central Register of Controlled Trials (CENTRAL) or the Cochrane COVID-19 Study Register (CCSR) are identified using Crowd and machine, helping to keep these important resources up to date.
Methods:
This presentation will provide a detailed look at the main component parts of the Evidence Pipeline applied at the specific review level (via S4M) and at the data repository level (CENTRAL and CCSR). It will also describe our evaluation work on each component part and on the system as a whole in comparison to traditional methods.
Results:
At the time of writing, the Evidence Pipeline has been used (via the Screen4Me workflow) in over 100 Cochrane reviews. The mean reduction in screening workload for author teams is 72% with a mean time to screening task completion of ten days per review. At the repository level, 99% of RCTs in CENTRAL have been identified by the Evidence Pipeline.
Conclusions:
The Cochrane Evidence Pipeline aims to expedite the study identification process for reviews and other evidence synthesis outputs, offering significant workload reduction for author teams within the current review production paradigm, but is also enabling a new paradigm to emerge—one that leverages economies of effort and scale to help create comprehensive, curated repos of studies for the EBHC communities who need it.
Patient, public, and/or healthcare consumer involvement:
The Evidence Pipeline uses Cochrane Crowd which is Cochrane’s citizen science platform. Anyone with an interest in health can join Cochrane Crowd. This collective effort has played a significant role in enabling the Cochrane Evidence Pipeline.