Endoscopic Stigmata: Recognition Lies in the Eye of the Beholder?
Bibliographic record
Abstract
The most widely accepted classification for endoscopic stigmata is the Forrest classification, which divides endoscopically visible lesions into 3 classes (ie, I, II, and III). The classification is most useful because it is based on risk of rebleeding in the absence of endoscopic therapy. But how good are we at identifying/classifying the stigmata on endoscopy? Data from the REASON Registry from Canada, which reviewed endoscopy data retrospectively, showed that high-risk stigmata were present in 47.8% of the patients on endoscopy, but less than two-thirds of these received endoscopic therapy. On the other hand, about 9.8% patients with low-risk stigmata received endoscopic therapy. More than one-quarter of the reports did not document and classify the stigmata. Similar studies by Lau et al and Bour et al have shown poor inter-observer agreement between experts for recognition of endoscopic stigmata. An anonymous online survey was designed and separate links to the survey were e-mailed to faculty and fellows. We divided the survey into 2 parts—stigmata recognition and stigmata therapy. Under stigmata recognition, a standard image of a lesion belonging to one class of the Forrest classification was presented. The images were reproduced from records of our own endoscopies or from standard online digital libraries of endoscopic stigmata. The images were chosen to represent a very standard imagery representative of the lesion. Under stigmata therapy, we used different labeled images of endoscopic lesions and asked faculty/fellows to choose how they would treat the lesion endoscopically, including an option to not treat. Total of 17 faculty members and 11 fellows participated in the survey. Our study brings forth a few vital observations:1.Based on our data, overall low-risk stigmata was misclassified as high risk about 14% of time by board-certified gastroenterologists. This would have inadvertently led to unnecessary endoscopic therapy. In the same vein, high-risk stigmata was misclassified as low risk 6% of the time, putting patients at risk for rebleeding from them.2.Overall inter-observer agreement in recognition of stigmata was very good (κ = .85). It was poorest for an adherent clot (κ = .52) and best for spurting hemorrhage (κ = 1).3.Overall inter-observer agreement in treatment of endoscopic stigmata was good (κ = .74). It was again poorest for an adherent clot (κ = .58) and best for spurting hemorrhage/clean-based ulcer and flat pigmented spot (κ = 1).4.The κ indexes were significantly better for faculty with >3 years of experience vs junior fellows (first- and second-year fellows) (senior faculty overall κ = .85 ± .09 and junior fellows .68 ± .16; 2-tailed P = .047).
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
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.011 | 0.063 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.003 | 0.005 |
| Scholarly communication | 0.004 | 0.015 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.005 | 0.006 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
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.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
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".