Multimodal deep learning of fundus abnormalities and traditional risk factors for cardiovascular risk prediction
importance 10/10
ai assisted
Retinal Imaging and Analysis cardiology stage: research
Problem
Cardiovascular disease, the leading global cause of death, needed better risk identification from accessible data.
Why it matters
CVD is the leading cause of death globally, so earlier risk detection could save lives.
What the AI did
A multimodal deep-learning model combined fundus photographs with clinical risk factors to predict cardiovascular disease risk.
Results
AUROC 0.781 (SMC) and 0.872 (UK Biobank external validation); hazard ratio 6.28 (95% CI 4.72-8.34) for incident CVD
Impact
Externally validated on UK Biobank; fundus photography proposed as a non-invasive predictive marker.
Record
| Venue | npj Digital Medicine |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | Yeong Chan Lee, Jiho Cha, Injeong Shim, Woong‐Yang Park, Se Woong Kang, Dong Hui Lim, Hong‐Hee Won |
| Citations | 107 (OpenAlex, at import time) |
| References | 40 |
| Model or method | multimodal deep learning model (not named) |
| Organization | Samsung Medical Center (development and internal validation) |
| Primary source | https://doi.org/10.1038/s41746-023-00748-4 |
| Open access | gold |
| Keywords | cardiovascular disease, fundus photography, multimodal deep learning, risk prediction, UK Biobank |
Importance rating
10
existential / global scale
Abstract states CVD is the leading cause of death globally; earlier risk detection from non-invasive fundus photos could affect millions of lives.