Jul 2026· Current Eye Research· Vol 51, pp. 1035 - 1043· 0 citations· 50 references
Medicine
Abstract
Abstract Purpose This study compared traditional statistical models with machine learning algorithms for predicting surgically induced astigmatism after cataract surgery, aiming to identify the most accurate and generalizable method among linear regression, regression trees, random forests, and neural networks. Methods Retrospective analysis was performed on 321 eyes (321 patients) undergoing phacoemulsification at a tertiary center. Data were standardized and surgically induced astigmatism was modeled using three vector-based outcomes: KEQ.post (keratometric equivalent power), KAST0.post (horizontal astigmatism component), and KAST45.post (oblique component). Preoperative variables served as predictors. Data were split into training (60%) and test (40%) sets. Four predictive models were developed and evaluated on unseen data. Performance was assessed using mean squared prediction error, and variable importance analyses identified key predictors. Results Linear regression achieved the best out-of-sample performance across all outcomes (e.g. KEQ.post mean squared prediction error = 0.043). Tree-based models performed slightly worse, while neural networks showed substantial overfitting with markedly higher test errors. Preoperative astigmatism and corneal radii were the strongest predictors. Conclusion KAST0 and KAST45 were not predictably modeled, whereas changes in KEQ.post were. Multivariable linear regression provided the most accurate and reliable predictions, while more complex machine-learning models (especially neural networks) overfit the limited dataset and offered no clinical advantage. The relationship between preoperative biometrics and postoperative equivalent power appears predominantly linear, though larger datasets may enhance machine-learning performance.
Surgeon seniority was not a significant determinant of surgically induced astigmatism after phacoemulsification cataract surgery, and machine-learning models based on preoperative clinical data provided only limited predictive value, particularly for vector astigmatism outcomes.
Denizcan Özizmirliler, C. Engin, Özlem Özkan et al.· BMC Ophthalmology· 0 citations
Objectives The magnitude and axis of surgically induced astigmatism depend on the sclera-corneal tunnel dimensions. Often, the incision depth is subjective, and standardization is rarely emphasized. This study attempts to evaluate a fixed scleral incision depth of 350 µm at 1-week post-op. The aim is to determine the SIA vector and centroid using the Pythagorean theorem and keratometric measurements. Materials and methods This study was conducted from January to July 2025 and recruited 42 patients at a tertiary health institution. All patients received scleral incisions to a depth of 350 µm after preoperative auto-keratometry measurements. Preoperative astigmatism was classified into WTR, ATR, and oblique types. Results The overall mean age was 67.14 (± 9.23) years, with 27 (64.29%) males and 15 (35.71%) females. A positive correlation (r = 0.93, 0.69) was found for pre- and post-operative keratometry. T-test for pre- and post-operative vertical keratometry revealed statistical significance (p < 0.0009). The mean SIA magnitude and axis were 2.05 (± 1.75) D and 77.94º (± 59.68º), respectively. The Wilcoxon signed-rank test showed statistical significance for SIA magnitude (p=0.0003). The centroid was at 22.3º. A 28.57% conversion rate from ATR to WTR was observed, with 71.43% remaining unchanged postoperatively. High prediction accuracy was observed for preoperative horizontal and vertical keratometry (R2 values of 0.93 and 0.74, respectively). Discussion This work adds a unique dimension to the literature by explicitly linking a fixed tunnel depth to the SIA vector and keratometric prediction. These results provide a platform for future research exploring scleral incision depth as a determinant of astigmatic outcomes. Conclusions This study finds a statistically significant difference in vertical keratometry with moderate SIA magnitude. Astigmatic axis transition was unchanged in most patients, likely due to a standardized scleral incision depth. Centroid demonstrated a minimal SIA magnitude on horizontal keratometry. An excellent postoperative keratometry prediction was determined.
P. Krishnacharya, Kedar Janardhan Prabhu, Shreya Nataraj et al.· Romanian Journal of Ophthalm...· 0 citations
AIM
To evaluate three-month visual and refractive outcomes after small incision lenticule extraction (SMILE) Pro with the VISUMAX 800 femtosecond laser for myopic astigmatism correction and to compare results with previously reported VISUMAX 800 and VisuMax 500 outcomes.
METHODS
Retrospective, observational clinical study. Records of 64 patients (128 eyes) who underwent SMILE Pro using the VISUMAX 800 laser (Carl Zeiss Meditec AG, Jena, Germany) between June and November 2024 were reviewed. Preoperative and 3-month postoperative data were analyzed in eyes with corrected distance visual acuity (CDVA) of 1.0 or better, including spherical equivalent (SE), refractive astigmatism, and vector analysis. Patients were classified into low (≤1.50 D) and high (>1.50 D) astigmatism subgroups.
RESULTS
The mean age of 64 patients (128 eyes) was 26.10±4.25y (19-36y). Attempted and achieved SE were strongly correlated (r=0.97, P<0.001). At 3mo, 36.7% and 87.5% of eyes were within ±0.50 and ±1.00 D of target refraction, respectively. Mean residual astigmatism was -0.56±0.27 D, lower in the low-astigmatism group (-0.39±0.23 D) than in the high group (-0.72±0.21 D; P<0.001). The mean absolute astigmatic angle of error was 4.47°±4.68°, with 97.6% ≤15°. No intraoperative or postoperative complications occurred.
CONCLUSION
SMILE Pro using the VISUMAX 800 femtosecond laser provides safe and predictable correction of myopic astigmatism with good axis alignment. However, the observed undercorrection in higher cylinders suggests that further optimization of astigmatic nomograms may improve refractive accuracy.
İ. İritaş, Z. Burke· International Journal of Oph...· 0 citations
Background/Objectives: High astigmatism remains challenging in keratorefractive lenticule extraction because small errors in centration or cyclotorsion may affect astigmatic correction. This study evaluated early refractive and astigmatic vector outcomes after small-incision lenticule extraction (SMILE) Pro using the VISUMAX 800 in eyes with high astigmatism. Methods: This prospective interventional cohort study was conducted at Cung Hong Son International Eye Hospital, Hanoi, Vietnam. Refractive outcomes, including uncorrected distance visual acuity (UDVA), corrected distance visual acuity (CDVA), manifest refraction, astigmatic vector analysis, and intraoperative and postoperative complications, were evaluated up to 3 months postoperatively. Results: A total of 160 eyes from 102 patients were included in the final analysis. The mean age was 23.7 ± 4.7 years, and 63.7% of participants were female. The mean preoperative spherical equivalent (SEQ) was −7.21 ± 2.03 D, and the mean preoperative refractive cylinder was −2.51 ± 0.56 D. Mean postoperative SEQ was −0.10 ± 0.38 D at 1 month and 0.02 ± 0.31 D at 3 months. At 3 months, 99% and 84% of eyes achieved UDVA of 0.8 decimal or better and 0.9 decimal or better, respectively. The mean residual refractive cylinder was −0.52 ± 0.28 D, with 69% and 98% of eyes achieving a residual cylinder ≤0.50 D and ≤1.00 D, respectively. All eyes had an angle of error within ±15°. The safety index was 1.01, and no eyes lost two or more lines of CDVA. Conclusions: SMILE Pro performed with the VISUMAX 800 provided favorable early 3-month refractive and astigmatic vector outcomes with a satisfactory safety profile in eyes with high astigmatism.
L. Tran, T. N. Tran, H. Cung et al.· Vision· 0 citations
This study successfully developed and validated an explainable RF-based machine learning model for the prediction of postoperative ACD in highly myopic cataract patients, which supports more reliable IOL power calculation and offers a practical tool for optimizing surgical planning in highly myopic eyes.
Yuyang Yang, Hao Cui, Jiajia Gao et al.· Frontiers in Cell and Develo...· 0 citations
Purpose To investigate the phenomenon of incision gape following subconjunctival oblique limbus incision (SCOLI) cataract surgery and to quantify its relationship with surgically induced astigmatism (SIA). Methods This study enrolled patients with regular corneal astigmatism ≥1.0 diopters (D) undergoing SCOLI. Anterior segment optical coherence tomography (AS-OCT) was used to measure the actual incision gape (ΔG). A theoretical model was developed to calculate the predicted incision gape (theoretical ΔG) based on preoperative keratometry and white-to-white (WTW) distance. The correlation between actual ΔG and SIA was analyzed using linear regression and a multivariate GEE model. Results Thirty-two eyes of 25 patients were analyzed. A significant positive correlation was observed between actual ΔG and SIA (R = 0.392, p = 0.026). In the multivariate GEE model, after adjusting for age (p = 0.195) and CCT (p = 0.070), actual ΔG remained a significant independent factor associated with SIA (B = 0.004; 95% CI, 0.002–0.006; p < 0.001). The theoretical ΔG showed a strong correlation with both the intraoperative ΔG (r = 0.877, p < 0.001) and the actual ΔG (r = 0.508, p = 0.003). Conclusion A macroscopic incision gape is a prevalent phenomenon in SCOLI and is a significant factor associated with SIA. The developed mathematical model provides a foundation for predicting astigmatic outcomes, highlighting the potential for controlling SIA by surgically modulating the incision gape (ΔG).
Xin Fan, P. Lai, Shaofeng Yang et al.· Clinical Ophthalmology· 0 citations