Predicting Postoperative Anterior Chamber Depth, Intraocular Lens Tilt, and Decentration Using an Internally...

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Data Science

Background:

• Precise intraocular lens (IOL) positioning is crucial for successful cataract surgery outcomes.
• IOL misalignment can cause refractive errors, astigmatism, and higher-order aberrations.
• Advanced IOLs require even more accurate placement for optimal visual acuity.

Purpose of the Study:

• To develop and validate machine learning models for predicting postoperative anterior chamber depth (ACD), IOL tilt, and decentration.
• To identify key preoperative biometric variables influencing IOL positioning.
• To assess the clinical applicability of predictive modeling in cataract surgery.

Main Methods:

• Prospective, single-center study involving 49 patients undergoing cataract surgery with a specific IOL type.
• Preoperative biometric data collected using swept-source optical coherence tomography.
• Partial least squares regression and random forest models used for prediction and validation.

Main Results:

• Machine learning models achieved low prediction errors: 0.07 mm for ACD, 1.17° for IOL tilt, and 0.02 mm for IOL decentration.
• Key predictors for ACD included preoperative ACD, axial length (AL), and lens tilt.
• Preoperative lens tilt, thickness, decentration, AL, and keratometry were significant predictors for IOL tilt and decentration.

Conclusions:

• Machine learning models demonstrate excellent predictability for postoperative ACD and IOL tilt.
• Good predictability was achieved for IOL decentration, supporting clinical use.
• Predictive modeling can optimize IOL positioning, enhancing visual outcomes in cataract surgery.