Routine blood tests and machine learning identify complications in high myopia.

Area of Science:

• Ophthalmology
• Medical Imaging
• Machine Learning

Background:

• High myopia poses significant risks for serious eye conditions like glaucoma and retinal detachment, leading to irreversible vision loss.
• Current screening methods rely on specialized imaging and limited specialist availability, hindering widespread population screening.

Purpose of the Study:

• To develop and validate a machine learning model that identifies individuals at increased risk of high myopia complications using routine blood test results.
• To assess the model's accuracy and effectiveness in real-world screening settings.

Main Methods:

• A machine learning model was developed and validated using data from 10,661 participants across multiple centers.
• The model's performance was further evaluated in independent cohorts and a large-scale community screening study involving 311,254 adults.

Main Results:

• The model demonstrated high accuracy across different centers (AUROC 0.9010-0.9649).
• It successfully flagged individuals diagnosed with high myopia complications in a prospective follow-up study.
• In community screening, the model improved complication detection rates among referred individuals (PPV=74%).

Conclusions:

• A scalable, blood-based machine learning approach can effectively identify individuals at high risk for high myopia complications.
• This method supports opportunistic screening in primary care and community settings, enabling earlier ophthalmic referral and potentially preventing vision loss.