MétaCan
Menu
Back to cohort
Record W2884485345 · doi:10.1111/jep.12993

What confidence should we have in <scp>GRADE</scp>?

2018· article· en· W2884485345 on OpenAlexaff
Mathew Mercuri, Brian S. Baigrie

Bibliographic record

VenueJournal of Evaluation in Clinical Practice · 2018
Typearticle
Languageen
FieldDecision Sciences
TopicMeta-analysis and systematic reviews
Canadian institutionsMcMaster UniversityUniversity of Toronto
Fundersnot available
KeywordsBayes' theoremConfidence intervalBayesian probabilityFrequentist inferenceSet (abstract data type)Evidence-based medicineQuality (philosophy)HierarchyMedicineComputer scienceBayesian inferenceArtificial intelligenceEpistemologyAlternative medicineInternal medicine

Abstract

fetched live from OpenAlex

RATIONALE, AIMS, AND OBJECTIVES: Confidence (or belief) that a therapy is effective is essential to practicing clinical medicine. GRADE, a popular framework for developing clinical recommendations, provides a means for assigning how much confidence one should have in a therapy's effect estimate. One's level of confidence (or "degree of belief") can also be modelled using Bayes theorem. In this paper, we look through both a GRADE and Bayesian lens to examine how one determines confidence in the effect estimate. METHODS: Philosophical examination. RESULTS: The GRADE framework uses a criteria-based method to assign a quality of evidence level. The criteria pertain mostly to considerations of methodological rigour, derived from a modified evidence-based medicine evidence hierarchy. The four levels of quality relate to the level of confidence one should have in the effect estimate. The Bayesian framework is not bound by a predetermined set of criteria. Bayes theorem shows how a rational agent adjusts confidence (ie, degree of belief) in the effect estimate on the basis of the available evidence. Such adjustments relate to the principles of incremental confirmation and evidence proportionism. Use of the Bayesian framework reveals some potential pitfalls in GRADE's criteria-based thinking on confidence that are out of step with our intuitions on evidence. CONCLUSIONS: A rational thinker uses all available evidence to formulate beliefs. The GRADE criteria seem to suggest that we discard some of that information when other, more favoured information (eg, derived from clinical trials) is available. The GRADE framework should strive to ensure that the whole evidence base is considered when determining confidence in the effect estimate. The incremental value of such evidence on determining confidence in the effect estimate should be assigned in a manner that is theoretically or empirically justified, such that confidence is proportional to the evidence, both for and against it.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Direct model labels (unvalidated)

Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.

Model armCategoriesStudy designConfidence
gemmaMetaresearch
Domain: Evaluation · Genre: Empirical
About the Canadian research system: no · About a Canadian topic: no
Theoretical or conceptuallow
gptMetaresearch
Domain: Evaluation · Genre: Commentary
About the Canadian research system: no · About a Canadian topic: no
Theoretical or conceptuallow
models agreeAgreement compares identical category sets and study designs across arms.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.664
metaresearch head score (Gemma)0.869
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesMetaresearch, Scholarly communication, Insufficient payload (model declined to judge)
Consensus categoriesMetaresearch, Insufficient payload (model declined to judge)
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Not applicable · Consensus signal: Not applicable
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.377
Threshold uncertainty score1.000

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.6640.869
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0030.001
Bibliometrics0.0010.002
Science and technology studies0.0000.000
Scholarly communication0.0010.004
Open science0.0010.000
Research integrity0.0000.001
Insufficient payload (model declined to judge)0.0050.002

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.916
GPT teacher head0.710
Teacher spread0.206 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Labeled directly by 2 models reading the full record.

Study designTheoretical or conceptual
DomainEvaluation
GenreEmpirical · Commentary

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations28
Published2018
Admission routes1
Has abstractyes

Explore more

Same venueJournal of Evaluation in Clinical PracticeSame topicMeta-analysis and systematic reviewsCategoryMetaresearchFrench-language works237,207