Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-2099
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-21T10:42:35.973601Z
status: published
lens: g_space
sector: health
headline: The Use of Artificial Intelligence to Predict Visual Outcome in Patients With Macular Pucker
dek: Purpose: To assess the current role of artificial intelligence (AI) to predict postoperative best-corrected visual acuity (BCVA) after macular pucker surgery using optical coherence tomography (OCT) data. Methods: A systematic search was conducted of PubMed, EMBASE, Web of Science, Scopus, and the Cochrane Library. Data extraction included study design, demographics, AI model characteristics, and performance metrics. Study quality was assessed using the Prediction Model Risk of Bias Assessment + Artificial Intel…
gain_title: Deep learning models applied to preoperative OCT scans predicted postoperative visual acuity after macular pucker surgery and outperformed human grading in the reviewed studies.
problem_title: (none)
trace_subject: (none)
gain_reading: Deep learning models applied to preoperative OCT scans predicted postoperative visual acuity after macular pucker surgery and outperformed human grading in the reviewed studies.
gain_evidence: ResNet-101 and Inception-ResNet-V2 outperformed human grading
problem_reading: (none)
problem_evidence: (none)
quick_read: A systematic review searched five major databases for studies that used artificial intelligence on optical coherence tomography data to predict best-corrected visual acuity after macular pucker surgery. It analyzed six studies covering 2728 patients aged 65 to 71, using convolutional neural networks and gradient boosting decision trees.

The findings suggest that imaging taken before surgery may help forecast how well a patient will see afterward, which could inform counseling and surgical planning. The authors stress that clinical use remains unproven until the models are validated more broadly, made interpretable, and tested in real time.
limitation: Clinical translation depends on model generalizability, interpretability, real-time application, and functional data, and the authors say multicenter validation is still needed.
tag: Evidence-backed gain
key_points: The review included six studies with 2728 patients aged 65 to 71. | ResNet-101 reached the highest area under the curve, at 0.97. | ResNet-101 and Inception-ResNet-V2 outperformed human grading. | Interpretability methods flagged ectopic inner foveal layers and an irregular inner nuclear layer-outer plexiform layer as key predictors.
rundown: The authors searched PubMed, EMBASE, Web of Science, Scopus, and the Cochrane Library, extracted study design, demographics, model characteristics, and performance metrics, and synthesized results narratively. Study quality was assessed with the Prediction Model Risk of Bias Assessment + Artificial Intelligence tool.

The models used convolutional neural networks and gradient boosting decision trees. Advanced architectures were reported to outperform simpler models and expert grading, and interpretability techniques pointed to specific retinal layer features as predictors of outcome.

The authors conclude that translation to practice depends on generalizability, interpretability, real-time application, and inclusion of functional data, and they call for multicenter validation.
sources:
- peer_reviewed | Journal of VitreoRetinal Diseases | https://doi.org/10.1177/24741264261495481 | 2026-10-20
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

c6ba8e22cbcb5c87f9579a91e335edcc937416d7ba2cf41e05a4d4c393fd908d

previous

0000000000000000000000000000000000000000000000000000000000000000