[Study on the design of refractive parameter scheme for SMILE surgery based on regression models for the correction...

Objective: To investigate the postoperative efficacy of a personalized refractive parameter optimization protocol constructed based on retrospective clinical data and regression models in femtosecond laser small-incision lenticule extraction (SMILE) for the correction of myopia and astigmatism. Methods: A cohort study was conducted. Prospectively and consecutively collected and enrolled 75 patients (148 eyes) with refractive errors who underwent SMILE guided by a regression model-based refractive parameter design protocol (nomogram optimized by Visulyze software) at the Refractive Center of West China Hospital, Sichuan University from August to December 2024, and designated them as the experimental group. Additionally, a control group was established using propensity score matching based on the data of the experimental group, with patients selected from those who received conventional SMILE between December 2022 and May 2023 and had complete 3-month postoperative follow-up data. All surgeries were performed using the VisuMax 500 kHz femtosecond laser system. Visual acuity and refractive power were examined using a standard logarithmic visual acuity chart, autorefractor, and phoropter. The primary outcome measures were refractive accuracy (proportion of eyes with spherical equivalent within ±0.50 D) and astigmatism correction (proportion of eyes with residual astigmatism ≤0.50 D) at 3 months postoperatively; the secondary outcome measures included efficacy, safety, and predictability. Results: The experimental group ultimately comprised 67 patients (133 eyes), yielding an attrition rate of 10.1%. There were 17 males and 50 females, with an age of 28.50 (24.00, 32.25) years. The control group included 67 matched patients (133 eyes), consisting of 15 males and 52 females, with an age of 29.00 (26.00, 31.25) years. The two groups were comparable. In terms of refractive accuracy, the proportion of eyes with spherical equivalent within ±0.50 D was 99.2% (132/133) in the experimental group and 88.7% (118/133) in the control group (P<0.001); the proportion of eyes with residual astigmatism ≤0.50 D was 99.2% (132/133) in the experimental group and 95.5% (129/133) in the control group (P=0.368). There were no statistically significant differences in efficacy and safety between the two groups (both P>0.05). For efficacy, the number of eyes with uncorrected visual acuity ≥20/20 at 3 months postoperatively was 129 (accounting for 97%) in the experimental group and 132 (accounting for 99%) in the control group. For safety, neither group had eyes with a loss of≥2 lines of best-corrected visual acuity. The experimental group showed significantly higher refractive predictability, with coefficients of determination (R²) for spherical equivalent and astigmatism of 0.978 8 and 0.865 3, respectively, compared with 0.950 3 and 0.726 4 in the control group. Conclusion: In this study, the personalized refractive parameter design protocol constructed based on retrospective clinical data and regression models significantly improves the accuracy of postoperative refractive prediction and the precision of astigmatism correction in femtosecond laser small-incision lenticule extraction (SMILE), and exhibits favorable efficacy and safety during the 3-month postoperative follow-up.