Nested Knowledge

Bringing Systematic Review to Life

User Tools

Site Tools

Trace:
wiki:autolit:screening:robot

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
wiki:autolit:screening:robot [2022/07/25 08:08]
kholub 		wiki:autolit:screening:robot [2024/06/10 19:42]
kevinkallmes removed
Line 1: Line 1:
-====== Robot Screener ======+===== Guidance on Robot Screener in Dual Two Pass Mode =====
  
-Robot Screener is an automated substitute for one of the human reviewers traditionally involved in dual screening.+The recommended practice is to **use Robot Screener as a reviewer in Abstract Screening** when in Dual Two Pass mode, and use fully manual Full Text screening.
  
-===== Enabling Robot Screener =====+   * Abstract Screening requires **50**  abstract adjudicated records, either advanced or excluded under Abstract Screening –> Adjudicate Screening, and **10**  advanced records 
 +      * These actions must be performed before Robot Screener can be turned on for Abstract Screening
  
-Robot Screener may be enabled from Screening Settings, under "Inclusion Modeling". Before Robot Screener can be enabled, the nest must: +  * Because fewer records are available, more intensive assessment of the full text is needed, and final decisions on inclusion are made at the Full Text stage, AutoLit does not currently offer Robot Screener at the Full Text Screening stage. 
- +      * You can keep Robot Screener toggled 'on' throughout your screening process, but once records are Advanced from Abstract Screening, each record must be screened by two independent screeners before proceeding to adjudication.
-  * Be configured for dual screening +
-  * And contain **50**  adjudicated //records //  (i.e. having a final screening decision) +
-  * And contain **5**  included records +
- +
-The latter two requirements help ensure a minimum model accuracy; however, the model will usually be suboptimal with this volume of training data, and typically improves as more records are screened. +
- +
-===== In Action ===== +
- +
-==== Which Records ==== +
- +
-Robot Screener automatically adds a reviewer-level screening decision to records with: +
- +
-  * Fewer than 2 reviewer-level screening decisions +
-  * No adjudicated screening decision +
- +
-Records it excludes will exclusively be assigned the exclusion reason ''Robot Excluded'' +
- +
-==== When ==== +
- +
-Robot Screener adds/updates its screening decisions when: +
- +
-  * Robot Screener is enabled +
-  * A new inclusion model is trained +
-  * New records are imported into your nest +
-  * 10 addditional records have been adjudicated, since the model was last trained +
- +
-As pointed out above, Robot Screener will not modify its decision on a record after that record has been adjudicated. Prior to adjudication, its decision on a record may be modified, reflecting more information available to it & improvement in its accuracy. +
- +
-===== Usage Guidance ===== +
- +
-Robot Screener, by accuracy, is not a replacement for a human reviewer. It should typically be used for reviews where budget, team size, or time available are restricted. Some additional usage tips: +
- +
-  * While Robot Screener may be enabled at 50 adjudicated / 5 included records, we typically advise not enabling it until a cross validation AUC of 0.7 or greater is achieved. +
-  * Avoid use of the "Auto-adjudicate" button, especially early on. The model (and human reviewer) should be subject to scrutiny. This will improve the quality of your review, reduce later reworks, and produce better training data for Robot Screener.+
  
  

Page Tools

Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 4.0 International
CC Attribution-Share Alike 4.0 International Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki