[Primary angle closure suspects: application of machine learning method for substantiation of close monitoring].

One of the priority areas in healthcare is the concept of predictive, preventive and personalized medicine, which is based on an individualized approach to the patient, including before the onset of diseases such as glaucoma.

Purpose: This study was conducted to substantiate the necessity of close monitoring of primary angle closure suspects (PACs) by comparing their clinical and anatomical parameters with those in normal eyes and in primary angle closure (PAC) before and after lens extraction (LE) or laser peripheral iridotomy (LPI).

Material and methods: This prospective study included 30 PACs patients. The comparison group consisted of 60 patients with PAC: 30 patients underwent LE with intraocular lens implantation, and 30 patients underwent LPI. Control group - 30 eyes without ophthalmic pathology. All subjects underwent swept-source optical coherence tomography (SS-OCT), including analysis of choroidal thickness in the macula, lens vault (LV), iris thickness and curvature (ICurv), and anterior chamber angle (ACA) profile. Machine learning methods were used, including data driven soft independent modelling of class analogies (DD-SIMCA).

Results: The parameters of PACs eyes occupied an intermediate position between those of PAC before treatment (according to DD-SIMCA classification) and normal eyes, but remained distinct from PAC eyes after treatment, falling outside the "safety zone" relative to normal values. Compared with the PAC group after LE, the PACs group had a shallower anterior chamber (2.60±0.13 mm vs. 3.63±0.199 mm, p=0.00), a narrower ACA profile (all p=0.00), a steeper iris (all p=0.00), lower uncorrected visual acuity (0.50±0.24 vs. 0.95±0.08, p=0.00), and a higher spherical equivalent (SE). Compared with PAC eyes after PLI, the PACs had greater LV (0.84±0.11 mm vs. 0.58±0.07 mm, p=0.00), higher intraocular pressure (19.7±0.8 mm Hg vs. 16.9±2.0 mm Hg, p=0.00), greater ICurv (all p<0.05), higher SE, and a narrower ACA profile.

Conclusion: Untreated PACs have significantly worse clinical and anatomical parameters, both in comparison with the norm and with PAC patients after treatment. This substantiates the need for closer monitoring of PACs.