Multimodal AI for Eye Health: Explainable, Early Anomaly Detection

Role: Research Fellow, ANU School of Computing.

This ongoing project develops multimodal machine-learning models for eye health, fusing structural and functional clinical data to detect anomalies earlier and more reliably than any single modality alone. The computer-science contribution centres on three ideas:

• Explainable AI (XAI) for clinical interpretation — models are built to be transparent and clinically meaningful, so a clinician can see why a prediction is made rather than trust a black box.
• Learning under the low-data clinical bottleneck — methods that stay robust when labelled medical data is scarce, a core constraint in real ophthalmology and precision-health settings.
• Earlier anomaly detection — surfacing subtle, pre-symptomatic signals to enable timely intervention.

The work is co-designed with vision researchers and clinicians at ANU JCSMR, aligning the modelling with real diagnostic workflows so the system augments expert judgement and moves toward co-designed, precision eye care.

A research paper and a grant application related to this work are currently under review; further detail will follow once outcome published.