Smartphone-Based Proactive Self-Screening for Ocular Surface Malignancies: A Nonrandomized Clinical Trial.

Area of Science:

• Ophthalmology
• Artificial Intelligence
• Public Health

Background:

• Ocular surface malignancies are often misdiagnosed due to subtle presentations and lack of screening tools.
• Delayed diagnosis can lead to vision loss and necessitate extensive surgical intervention.

Purpose of the Study:

• To develop and validate a smartphone-based AI system for proactive self-screening of ocular surface malignancies.
• To enable early detection and triage of high-risk cases in the general population.

Main Methods:

• A deep learning model was trained on slitlamp images and optimized for smartphone photography.
• The CaptureTumor mobile application provided AI-guided imaging and immediate risk stratification.
• A nonrandomized clinical trial involved multimedia outreach and at-home self-screening.

Main Results:

• The smartphone system achieved an AUC of 0.905, comparable to slitlamp performance.
• Population-level screening demonstrated high accuracy (AUC 0.977, sensitivity 89.3%, specificity 95.9%).
• 20 malignancies were confirmed, with 95% newly diagnosed, and no enucleations were required.

Conclusions:

• Smartphone-enabled AI screening is a scalable, accessible, and effective strategy for early detection of rare ocular malignancies.
• This mobile health model addresses gaps in equitable care for vision-threatening diseases.