Choosing the formula for calculating optical power of intraocular lens on "short" eyes using the potential of artificial intelligence

Abstract

Objective: to carry out a comparative analysis of IOL fitting accuracy in patients with eye length less than 22.0 mm using Barrett Universal II, Kane, Hoffer Q and artificial intelligence formulas. Material and methods. The results of 88 phacoemulsification cataract operations with implantation of monofocal lOLs were analyzed. Preoperative biometry and IOL calculations were performed on the IOL Master 700. The accuracy of IOL fitting was also determined using the LensCalc program, which is based on artificial intelligence (DecisionTreeClassifier). Results. The eye length of the patients ranged from 19.8 to 22 mm. The prediction of the target refraction hit was the most accurate when using Barrett Universal II formulas as opposed to Hoffer Q (Z=2.12; p=0.034). The mean value of the target refraction hit error was not different when using the Barrett Universal II formula compared to the Kane formula (p>0.05). Using artificial intelligence it was found that higher accuracy in the results of IOL optical power determination was achieved when using the Barrett Universal II formula. Conclusion. Based on a comparative analysis of the study results and assessment of IOL fitting accuracy using artificial intelligence, it was found that the Barrett Universal II formula (4th generation) is more accurate in determining the optical power of lOLs in "short" eyes than the Hoffer Q formula (3rd generation). The results obtained using the Barrett Universal II formula, in contrast to the Hoffer Q formula, do not differ from the data calculated using the Kane formula (5th generation), which, according to the results of most studies, is the most accurate IOL selection formula at present (p>0.05).