Artificial intelligence assistance narrows the experience gap in endoscopic reporting of gastric lesions: a prospective clinical trial
Bibliographic record
Abstract
BACKGROUND: High-quality endoscopic reports are crucial for clinical decision-making in gastric lesions. However, manually generated reports often suffer from inconsistencies, resulting in incomplete lesion documentation and insufficient descriptions. The objective is to validate an artificial intelligence (AI) reporting system for improving the quality of gastric lesion documentation with a particular focus on narrowing the experience gap among endoscopists. METHODS: We retrospectively collected gastric images to develop a reporting system for gastric lesions. The system comprises several deep learning models for lesion detection, classification, and feature recognition, and was validated using 276 video clips. Both retrospective and prospective case validations were conducted to evaluate the system's clinical effectiveness, comparing the completeness of reports with and without AI assistance, particularly in describing features of suspicious lesions. RESULTS: In video validation, the system identified 99.27% (274/276) of lesions, with 88.97% accuracy in neoplasm detection, achieving 93.75% sensitivity and 86.98% specificity. Retrospective analyses showed that junior endoscopists using AI reported significantly more complete lesion documentation than original reports (74.75% vs 58.08%, P < 0.001). In the prospective trial, the AI-assisted group demonstrated superior lesion reporting completeness (86.50% vs 75.11%, P = 0.001). Notably, the AI group reported all 18 identified suspicious neoplasms, while the routine group reported only 16. CONCLUSIONS: This study confirms the efficacy of an AI reporting system in documenting gastric focal lesions, highlighting significant improvements in the completeness and accuracy of reports generated by endoscopists.
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 | 0.012 | 0.015 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.002 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
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".