Сonvolutional neural network trained on an ultra-small sample to identify non-modifiable cardiovascular risk factors (sex and age) by digital fundus photographs

Aim. To evaluate the effectiveness of a convolutional neural network trained on an ultrasmall sample for identifying non-modifiable cardiovascular risk factors (sex and age) by digital fundus photographs.

Material and methods. The EfficientNet B3 architecture, pretrained on the ImageNet database, was used. The study was conducted on a proprietary dataset containing digital fundus photographs and patient demographic data, divided into training (227 photos) and test (131 photos) samples. To determine the accuracy of age prediction, the mean absolute error (MAE), the coefficient of determination (R²), and the BlandAltman plots were evaluated. For sex prediction, sensitivity, specificity, positive and negative predictive values, and the area under the ROC curve were assessed.

Results. The MAE for age was 6,04 (95% confidence interval (CI): 5,11-7,11), while R² — 0,638 (95% CI: 0,486-0,759). The area under the ROC curve for sex prediction was 0,79 (95% CI: 0,70-0,87). Sensitivity, specificity, negative and positive predictive values, and balanced accuracy (at a probability threshold of 0,5) were 88, 58,1, 81,8, 70,1, and 73,2%, respectively.

Conclusion. The obtained results demonstrate high accuracy in sex determination and moderate accuracy in age determination, indicating that acceptable results can be achieved even with a very small dataset.

1. Visseren FLJ, Mach F, Smulders YM, et al. ESC Guidelines on cardiovascular disease prevention in clinical practice. Eur Heart J. 2021;42:3227-3337. doi:10.1093/eurheartj/ehab484

2. Damen JAAG, Hooft L, Schuit E, et al. Prediction models for cardiovascular disease risk in the general population: systematic review. BMJ. 2016;353:i2416. doi:10.1136/bmj.i2416

3. van Trier TJ, Snaterse M, Boekholdt SM, et al. Validation of Systematic Coronary Risk Evaluation 2 (SCORE2) and SCORE2-Older Persons in the EPIC-Norfolk prospective population cohort. Eur J Prev Cardiol. 2024;31:182-189. doi:10.1093/eurjpc/zwad318

4. Poplin R, Varadarajan AV, Blumer K, et al. Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning. Nat Biomed Eng. 2018;2:158-164. doi:10.1038/s41551-018-0195-0

5. Vishwakarma P, Tripathi, Singh BP, et al. Segmentation and Detection of Brain Tumors using EfficientNetB3 Model. In: 2023 2nd International Conference on Automation, Computing and Renewable Systems (ICACRS). Pudukkottai, India; 2023:1463-1470. doi:10.1109/ICACRS58579.2023.10404936

6. Kumar A, Nelson L. Deep Learning-based Blood Cell Classification using EfficientNetB3 Architecture. In: 2025 6th International Conference on Mobile Computing and Sustainable Informatics (ICMCSI). Goathgaun, Nepal; 2025:954-960. doi:10.1109/ICMCSI64620.2025.10883188

7. Alshoraihy A. EfficientNetB3 in Leukemia Detection: Advancements in Medical Imaging Analysis. Medinformatics. 2025. doi:10.47852/bonviewMEDIN52023293

8. Weiss K, Khoshgoftaar TM, Wang DD. A survey of transfer learning. J Big Data. 2016;3:9. doi:10.1186/s40537-016-0043-6

9. Ganaie MA, Hu M, Malik AK, et al. Ensemble deep learning: A review. Eng Appl Artif Intell. 2022;115:105151. doi:10.1016/j.engappai.2022.105151

10. Xu M, Yoon S, Fuentes A, et al. A comprehensive survey of image augmentation techniques for deep learning. Pattern Recognit. 2023;109347. doi:10.1016/j.patcog.2023.109347

11. Dieck S, Ibarra M, Moghul I, et al. Factors in Color Fundus Photographs That Can Be Used by Humans to Determine Sex of Individuals. Trans Vis Sci Tech. 2020;9(7):8. doi:10.1167/tvst.9.7.8

12. Korot E, Pontikos N, Liu X, et al. Predicting sex from retinal fundus photographs using automated deep learning. Sci Rep. 2021;11:10286. doi:10.1038/s41598-021-89743-x

13. Kim YD, Noh KJ, Byun SJ, et al. Effects of Hypertension, Diabetes, and Smoking on Age and Sex Prediction from Retinal Fundus Images. Sci Rep. 2020;10:4623. doi:10.1038/s41598-020-61519-9

14. Betzler BK, Yang HHS, Thakur S, et al. Gender Prediction for a Multiethnic Population via Deep Learning Across Different Retinal Fundus Photograph Fields: Retrospective Cross-sectional Study. JMIR Med Inform. 2021;9(6):e25165. doi:10.2196/25165

15. Taha AM, Zarandah QMM, Abu-Nasser BS, et al. Gender Prediction from Retinal Fundus Using Deep Learning. Int J Acad Inf Syst Res. 2022;6(5):57-63.

16. Gerrits N, Elen B, Craenendonck TV, et al. Age and sex affect deep learning prediction of cardiometabolic risk factors from retinal images. Sci Rep. 2020;10:9432. doi:10.1038/s41598-020-65794-4

17. Khan NC, Perera C, Dow ER, et al. Predicting Systemic Health Features from Retinal Fundus Images Using Transfer-Learning-Based Artificial Intelligence Models. Diagnostics. 2022;12(7):1714. doi:10.3390/diagnostics12071714

18. Berk A, Ozturan G, Delavari P, et al. Learning from small data: Classifying sex from retinal images via deep learning. PLoS ONE. 2023;18(7):e0289211. doi:10.1371/journal.pone.0289211