Assessment Procedure Guidance

From QIBA Wiki
Jump to: navigation, search

Current Document : QIBA Assessment Procedure Guidance 2019.03.05

This Guidance supplements the QIBA Profile Template and the Claim Guidance by describing the process of developing appropriate procedures to assess profile requirements, particularly those related to the statistical assumptions underlying the claim (e.g. the repeatability coefficient for the volumes generated by the tumor segmentation software).


Suggestions

Submit suggested improvements/gaps in this document by adding bullets here:

  • Section 2 would benefit from more examples from other biomarker groups
  • In the Site Assessment Datasets section:
  • Consider referencing the Metrology papers for more detail. Perhaps some discussion of "study design" for a test-retest dataset on humans.
  • Consider expressing some material in this subsection as a table of characteristics and pros and cons for different types of datasets. Precision, Bias, Linearity and Regression Slope assessments will prioritize some of the characteristics differently, leading to different conclusions.
  • Consider laying out some of the relevant considerations/rationales for the mentioned "judgement call" and perhaps cite some papers that have made one choice or another as instructive examples.

They will be reviewed by the Process Committee and addressed in upcoming versions.

Past Versions