Table 2 Summary of studies in the current literature (cardiovascular disease)
From: Prognostic potentials of AI in ophthalmology: systemic disease forecasting via retinal imaging
Author | Study year | Key findings | Retinal biomarker | Dataset | Total No. of events/ participants | Adjusted variable | Study design |
|---|---|---|---|---|---|---|---|
Rim et al. [16] | - | Reti-CAC is a reliable and effective alternative to CT scan-measured coronary artery calcification (CAC) for predicting cardiovascular events. Adding Reti-CAC to the PCE enhances cardiovascular risk stratification, evidenced by a continuous net reclassification index of 0.261. | Reti-CAC score | Severance Main Hospital, Seoul, South Korea CMERCÂHI (South Korea) SEED (Singapore) UK Biobank (UK) | 527 (CMERC-HI) 33 (6.3%) of 527 had CVD events during the 5-year follow-up 8,551 (SEED) 310 (3.6%) of 8551 had CVD events 47,679 (UK Biobank) 337 (0.7%) of 47679 had CVD events | Age Gender Hypertension Dyslipidaemia Diabetes Smoking status BMI Fasting glucose level | Retrospective cohort study |
Tseng et al. [17] | May 2021 | Reti-CVD, derived from retinal photographs, indicates a higher CVD risk, and serves as an effective tool for identifying individuals with a ≥ 10% 10-year CVD risk, including those in the borderline-QRISK3 category. It accurately identifies high-risk individuals, with a 10-year risk of 13.1% (CI: 11.7%–14.6%), enhancing early intervention and refining risk assessments, especially for those with a 10-year risk of 7.5%–10%. | Reti-CVD score | UK Biobank | 48,260 (6.3% of 48,260 had CVD events during the 11-year follow-up) | Age Gender Smoking status Body mass index Hypertension Total cholesterol level Diabetes mellitus status | Retrospective cohort study |
Chang et al. [18] | Between January 2005 and December 2017 | DL-FAS, using retinal fundus images, predicts CVD death risk, adding value beyond the Framingham risk score (FRS). It indicates atherosclerosis severity with a DL-FAS > 0.66 marking a significantly higher CVD death risk (HR 8.33) compared to DL-FAS < 0.33. Enhancing FRS models, DL-FAS raises concordance by 0.0266 and stands as an independent CVD death predictor. | DL-FAS | HPC-SNUH | 32,227 | Age Gender BMI FRS risk level Smoking status Alcohol consumption Exercise Diabetes status Hypertension Dyslipidaemia | Retrospective cohort study |
Cheung et al. [19] | January 2005 and December 2016 | Retinal-vessel calibre measurements from DL models correlate with CVD risk, matching or surpassing expert graders in predictive accuracy. These models achieve high precision in retinal-vessel measurement, with intraclass correlation coefficients of 0.82 to 0.95 compared to expert assessments. | SIVA-DLS | SEED Dataset | 59,191 | Age Gender Ethnicity MABP BMI Total cholesterol level HbA1c Smoking status | Retrospective cohort study |
Poplin et al. [20] | UK Biobank –2006 to 2010 EyePACS – 2007 to 2015 | DL models using retinal fundus images accurately predict the onset of MACE within 5 years and quantify previously elusive cardiovascular risk factors (age, gender, smoking status, blood pressure). Predictive accuracies include age with mean absolute error 3.26 years, gender (AUC: 0.97), smoking status (AUC: 0.71), systolic blood pressure with mean absolute error 11.23 mmHg, and MACE (AUC: 0.70). | Retinal fundus images | UK Biobank EyePACs Dataset | 12,026 patients, 91 with prior cardiac events were excluded. In the remaining 11,835 patients, 105 experienced MACE within 5 years post-retinal imaging. | Age Gender Smoking status HbA1c Blood pressure BMI Major adverse Cardiovascular events | Observational cohort study |