AI Streamlines GA Trial Screening, Study Finds

A clinic-based algorithm reduced screen failures and showed close agreement with manual measurements in data presented at ARVO 2026.

Jim Gallagher, senior managing editorJim Gallagher, senior managing editor

Retinal Physician May 4, 2026

Clinical Report: AI Streamlines GA Trial Screening, Study Finds

Overview

A clinic-based AI algorithm significantly reduced screen failures in geographic atrophy (GA) trial prescreening and demonstrated close agreement with manual measurements. The study suggests that AI tools can enhance trial efficiency and reduce patient burden.

Background

Accurate measurement of GA lesion area is crucial for clinical trial enrollment, yet many sites struggle with practical methods for determining eligibility. High screen failure rates can burden patients and investigators alike. The introduction of AI tools in clinical settings may address these challenges by streamlining the prescreening process.

Data Highlights

Measure	AI	Manual
Mean GA Area (mm²)	8.98	9.39
False Positive Rate	12%	N/A
False Negative Rate	1%	N/A
Screen Failure Rate	16%	25%

Key Findings

AI-assisted prescreening identified 80 eligible eyes out of 107 evaluated.
Manual confirmation validated 67 of the 80 prescreened eyes as eligible.
The mean GA area measured by AI was 8.98 mm², compared to 9.39 mm² by manual planimetry.
The screen failure rate was reduced from 25% to 16% with AI assistance.
False-positive and false-negative rates were 12% and 1%, respectively.

Clinical Implications

The integration of AI tools in clinical trial prescreening may lead to more efficient patient selection and reduced screening burdens. This advancement could enhance the overall effectiveness of clinical trials in ophthalmology.

Conclusion

The findings from this study highlight the potential of AI to improve the prescreening process for GA trials, suggesting a shift towards more efficient and accurate eligibility assessments.

PentaVision

Retinal Physician