Artificial intelligence in ophthalmology III: systemic disease prediction

For decades, researchers have sought to measure retinal changes in identifying ocular biomarkers for systemic diseases, newly termed oculomics. As big multimodal ocular image datasets are increasingly available and complex, artificial intelligence (AI)–based ocular image analysis is promising in providing a noninvasive tool for the prediction of various systemic diseases, through the evaluation of risk factors, retinal features, and biomarkers. Most existing AI algorithms focus on retinal photography–based systemic disease detection. Other retinal imaging modalities such as optical coherence tomography (OCT) and OCT angiography hold great potential for further research. It is important to visualize the specific ocular features identified by AI-based ocular image analysis for systemic diseases and provide a more comprehensive understanding of eye-body relations. Multitask and multimodal AI algorithms should be established to improve the accuracy of systemic disease identification (e.g., concurrent systemic diseases and eye pathologies). Combining the use of portable devices and telemedicine and AI-based ocular image analysis can potentially facilitate systemic disease screening in the general population and less-developed areas, enabling more patients to be diagnosed and treated at an earlier stage.