The use of artificial intelligence in measuring intraocular pressure

Relevance High-tech ophthalmological care is often unavailable in areas with low economic opportunities. One of the important aspects of an ophthalmological examination is measuring intraocular pressure (IOP) [1]. The basic publicly available test is Maklakov tonometry. However, even with it, there are certain difficulties: in particular, non-compliance with the standards for the manufacture of measuring rulers for impressions [2–4]. A simple way to assess the IOP level using a smartphone can be a telegram-bot, which allows you to estimate the diameter of impressions, eliminating measurement errors. Objective Comparative assessment of the accuracy of artificial intelligence technologies in fixed-force applanation tonometry. Material and methods The study involved 290 patients (576 eyes) for Maklakov tonometry with weights weighing 5; 10 and 15 g using a modified elastotonometry technique [5]. The quality analysis of the study and the measurement of the impressions were carried out by three independent experts. In parallel, the results were assessed using artificial intelligence. Results The data analysis by the neural network was comparable with the measurement data of three experts, but there was an underestimation of the measurement results by artificial intelligence by an average of 0.27 (–3.81; 4.35) mm Hg. The correlation coefficient for all impressions was 98.3%. Conclusion Artificial intelligence showed high accuracy and reproducibility of measurements. Key words: glaucoma; intraocular pressure; tonometry; artificial intelligence