Predictive modeling of glaucomatous optic neuropathy progression rate using machine learning methods

PURPOSE. To develop a method for individualized prediction of the rate of progression in moderate and advanced primary open-angle glaucoma (POAG).

METHODS. The study included patients with confirmed moderate and advanced POAG who had been under observation for at least 36 months. Prediction of the rate of visual function loss was performed using modern machine learning methods, specifically Ranked PLS-DA, which is highly resistant to multicollinearity and allows for the ordered nature of classes to be taken into account. Two sets of input data were considered: a complete set of 34 variables and an optimized set of 20 variables, including demographic, functional, structural, and vascular indicators. For model optimization and validation, a test dataset was artificially generated using Procrustes Cross-Validation (PCV) method. Model performance was evaluated using specific metrics: sensitivity, specificity, total efficiency (TEFF), and area under the ROC curve (AUC).

RESULTS. The optimized set of variables improved model sensitivity (0.93 vs 0.78) while maintaining high specificity (0.78). Total efficiency on the test dataset was 0.77 for the reduced set, with an AUC of 0.9. The model not only distinguished patients with fast, moderate, and slow progression rates but also identified «borderline» cases requiring closer monitoring. Analysis of individual variable contributions revealed key predictors influencing prognostic accuracy: age, retinal nerve fiber layer thickness and ganglion cell complex thickness, peripapillary vessel density, and parafoveal retinal thickness. These findings underscore the importance of a comprehensive approach to assessing the risk of irreversible changes in visual functions.

CONCLUSION. The developed Ranked PLS-DA model demonstrated high effectiveness in stratifying patients with moderate and advanced glaucoma according to progression rate. The model may serve as a reliable basis for individualized monitoring and therapy in routine clinical practice.

1. Vision Loss Expert Group of the Global Burden of Disease Study; GBD 2019 Blindness and Vision Impairment Collaborators. Global estimates on the number of people blind or visually impaired by glaucoma: A meta-analysis from 2000 to 2020. Eye 2024; 38:2036-2046. https://doi.org/10.1038/s41433-024-02995-5

2. European Glaucoma Society. Terminology and guidelines for glaucoma. 5th edition. Br J Ophthalmol 2021; 105(Suppl 1):1-169. https://doi.org/10.1136/bjophthalmol-2021-egsguidelines

3. Hu R, Racette L, Chen KS, Johnson CA. Functional assessment of glaucoma: Uncovering progression. Surv Ophthalmol 2020; 65(6):639-661. https://doi.org/10.1016/j.survophthal.2020.04.004

4. Miki A. Assessment of Structural Glaucoma Progression. J Curr Glaucoma Pract 2012; 6(2):62-67. https://doi.org/10.5005/jp-journals-10008-1108

5. Medeiros FA, Zangwill LM, Girkin CA, Liebmann JM, Weinreb RN. Combining structural and functional measurements to improve estimates of rates of glaucomatous progression. Am J Ophthalmol 2012; 153(6):1197-205.e1. https://doi.org/10.1016/j.ajo.2011.11.015

6. Yang WYL, Wong HJ, Fu CE, Rojas-Carabali W, Agrawal R, Ang BCH. Artificial intelligence in the prediction of glaucoma development and progression: A systematic review. Surv Ophthalmol 2025; S0039-6257(25)00105-5. https://doi.org/10.1016/j.survophthal.2025.06.006

7. Shuldiner SR, Boland MV, Ramulu PY, De Moraes CG, Elze T, Myers J, et al. Predicting eyes at risk for rapid glaucoma progression based on an initial visual field test using machine learning. PLoS One 2021; 16(4):e0249856. https://doi.org/10.1371/journal.pone.0249856

8. Mariottoni EB, Datta S, Shigueoka LS, Jammal AA, Tavares IM, Henao R, et al. Deep Learning-Assisted Detection of Glaucoma Progression in Spectral-Domain OCT. Ophthalmol Glaucoma 2023; 6(3):228-238. https://doi.org/10.1016/j.ogla.2022.11.004

9. Kurysheva N.I., Kim V.Yu., Kim V.E., Laver A.B. Lamina cribrosa curvature index and its relationship with morphofunctional and microcirculatory disorders in glaucoma. National Journal glaucoma 2023; 22(3):15-25.

10. Rao HL, Kumbar T, Kumar A, Chandrasekhar G, Rao G, Garudadri CS. Agreement between event-based and trend-based glaucoma progression analyses. Eye 2013; 27:803-808. https://doi.org/10.1038/eye.2013.77

11. Holló G, Zhou Q. Evaluation of Retinal Nerve Fiber Layer Thickness and Ganglion Cell Complex Progression Rates in Healthy, Ocular Hypertensive, and Glaucoma Eyes With the Avanti RTVue-XR Optical Coherence Tomograph Based on 5-Year Follow-up. J Glaucoma 2016; 25(10):e905-e909. https://doi.org/10.1097/IJG.0000000000000410

12. Swaminathan SS, Jammal AA, Berchuck SI, Medeiros FA. Rapid initial OCT RNFL thinning is predictive of faster visual field loss during extended follow-up in glaucoma. Am J Ophthalmol 2021; 229:100-107. https://doi.org/10.1016/j.ajo.2021.03.019

13. Pomerantsev A, Rodionova O. A new method for soft probabilistic discrimination of ranked classes. Journal of Mathematical Analysis and Applications 2025; 512(3):456-470. https://doi.org/10.2139/ssrn.5128628

14. Kucheryavskiy S, Rodionova O, Pomerantsev A. Procrustes crossvalidation of multivariate regression models. Anal Chim Acta 2023; 1255:341096. https://doi.org/10.1016/j.aca.2023.341096

15. Shoji T, Zangwill LM, Akagi T, Saunders LJ, Yarmohammadi A, Manalastas PIC, et al. Progressive Macula Vessel Density Loss in Primary Open-Angle Glaucoma: A Longitudinal Study. Am J Ophthalmol 2017; 182:107-117. https://doi.org/10.1016/j.ajo.2017.07.011

16. Kurysheva NI. The role of retinal microcirculation disorders in the progression of glaucomatous optic neuropathy. Russian Annals of Ophthalmology 2020; 136(4):57-65. https://doi.org/10.17116/oftalma202013604157

17. Dixit A, Yohannan J, Boland MV. Assessing Glaucoma Progression Using Machine Learning Trained on Longitudinal Visual Field and Clinical Data. Ophthalmology 2021; 128(7):1016-1026. https://doi.org/10.1016/j.ophtha.2020.12.020

18. Li F, Su Y, Lin F, Li Z, Song Y, Nie S, et al. A deep-learning system predicts glaucoma incidence and progression using retinal photographs. J Clin Invest 2022; 132(11):e157968. https://doi.org/10.1172/JCI157968

19. Hussain S, Chua J, Wong D, Lo J, Kadziauskiene A, Asoklis R, et al. Predicting glaucoma progression using deep learning framework guided by generative algorithm. Sci Rep 2023; 13(1):19960. https://doi.org/10.1038/s41598-023-46253-2

20. Kurysheva NI, Lepeshkina LV, Shatalova EO. Comparative study of factors associated with the progression of primary open-angle and angle-closure glaucoma. Russian Annals of Ophthalmology 2020; 136(2):64-72. https://doi.org/10.17116/oftalma202013602164

21. Jammal AA, Berchuck SI, Thompson AC, Costa VP, Medeiros FA. The effect of age on increasing susceptibility to retinal nerve fiber layer loss in glaucoma. Invest Ophthalmol Vis Sci 2020; 61(13):8. https://doi.org/10.1167/iovs.61.13.8

22. Anraku A, Enomoto N, Takeyama A, et al. Baseline thickness of macular ganglion cell complex predicts progression of visual field loss. Graefes Arch Clin Exp Ophthalmol 2014; 252(1):109-115. https://doi.org/10.1007/s00417-013-2527-9

23. Holló G. Optical coherence tomography angiography in glaucoma: analysis of the vessel density-visual field sensitivity relationship. Ann Transl Med 2020; 8(18):1203. https://doi.org/10.21037/atm.2020.03.187

24. Moghimi S, Zangwill LM, Penteado RC, Hasenstab K, Ghahari E, Hou H, et al. Macular and optic nerve head vessel density and progressive retinal nerve fiber layer loss in glaucoma. Ophthalmology 2018; 125(11):1720-1728. https://doi.org/10.1016/j.ophtha.2018.05.006

25. Moghimi S, Bowd C, Zangwill LM, Penteado RC, Hasenstab K, Hou H, et al. Measurement Floors and Dynamic Ranges of OCT and OCT Angiography in Glaucoma. Ophthalmology 2019; 126(7):980-988. https://doi.org/10.1016/j.ophtha.2019.03.003

26. Shen R, Chan LKY, Yip ACW, Chan PP. Applications of optical coherence tomography angiography in glaucoma: current status and future directions. Front Med (Lausanne) 2024; 11:1428850. https://doi.org/10.3389/fmed.2024.1428850

27. Onishi AC, Treister AD, Nesper PL, Fawzi AA, Anchala AR. Parafoveal vessel changes in primary open-angle glaucoma and normal-tension glaucoma using optical coherence tomography angiography. Clin Ophthalmol 2019; 13:1935-1945. https://doi.org/10.2147/OPTH.S206288

28. Kurysheva NI, Ponomareva SI, Maslova EV, Kim VE, Rodionova OYe, Pomerantsev AL. Predictive modeling of glaucomatous optic neuropathy progression rate in patients with newly diagnosed early primary open-angle glaucoma. Russian Annals of Ophthalmology 2025; 141(2):22-29. https://doi.org/10.17116/oftalma202514102122