AI-Driven IOL Prediction Using SS-OCT Biometry

N/ANot Yet RecruitingNCT07733154

The Hong Kong Polytechnic University2,500 enrolled

Overview

This prospective, multicenter, non-interventional observational study aims to evaluate the agreement between a novel swept-source optical coherence tomography (SS-OCT) biometer and the IOLMaster 700 for ocular biometric measurements and to develop and validate an artificial intelligence (AI)-driven intraocular lens (IOL) power prediction model. The primary research questions are: Does the novel SS-OCT biometer provide ocular biometric measurements that are comparable to those obtained using the IOLMaster 700? Can an AI-driven IOL power prediction model developed from multicenter biometric, imaging, surgical, and postoperative refractive data improve the accuracy of postoperative refractive prediction? Adult patients with age-related cataract who are scheduled to undergo routine phacoemulsification with posterior chamber IOL implantation will be consecutively recruited from participating study centers. Participants will undergo routine preoperative clinical assessments together with additional non-contact biometric measurements using the novel SS-OCT biometer before cataract surgery. Cataract surgery and postoperative care will follow standard clinical practice and will not be influenced by the study. Postoperative refractive and visual outcomes at approximately one month after surgery will be collected for analysis. The AI model and research device measurements will be used for research purposes only and will not be used to guide clinical decision-making during the study.

Study Type

OBSERVATIONAL

Enrollment

2,500

Conditions

Cataract

Interventions

Eligibility

Sex: ALLMin age: 18 Years

Medical Language ↔ Plain English

Inclusion Criteria: 1. Aged 18 years or above. 2. Diagnosed with age-related cataract. 3. Scheduled for routine phacoemulsification with posterior chamber intraocular lens implantation. 4. Able to cooperate with preoperative ocular biometric measurements and postoperative assessments. 5. Able to attend the planned 1-month postoperative follow-up visit. 6. Able to understand the study information and provide written informed consent. Exclusion Criteria: 1. Ocular conditions that may significantly affect biometric measurements or postoperative refractive outcomes, including but not limited to severe corneal disease, severe ocular surface disease, advanced glaucoma, retinal detachment, severe maculopathy, significant optic nerve disease, or ocular trauma. 2. History of previous intraocular surgery in the study eye, except minor superficial ocular procedures that are not expected to affect ocular biometry or refractive outcomes. 3. Uncontrolled severe systemic disease that, in the opinion of the investigator or treating clinician, makes the patient unsuitable for routine cataract surgery or study participation. 4. Cognitive impairment, severe mental disorder, or other condition that prevents the participant from understanding the study or providing valid informed consent. 5. Poor cooperation that prevents reliable ocular biometric measurement or follow-up assessment. 6. Current participation in another interventional clinical study that may affect the outcomes of this study.

Locations (3)

Royal Adelaide Hospital (tentative)

Adelaide, Australia

Wuhan Purui Ophthalmology Hospital

Wuhan, Hebei, China

Hong Kong Eye Hospital

Kowloon City, Hong Kong

Outcomes

Primary Outcomes

Agreement of ocular biometric parameters between Novel SS-OCT Biometer and IOLMaster 700

Bland-Altman 95% limits of agreement, mean difference, intraclass correlation coefficient (ICC) for six key biometric parameters: axial length, keratometry, anterior chamber depth, lens thickness, central corneal thickness, white-to-white diameter

Time frame: Preoperatively (single measurement visit before cataract surgery)

Secondary Outcomes

Measurement success rate of two biometers

Percentage of valid, analyzable scans obtained from Novel SS-OCT Biometer vs IOLMaster 700

Time frame: Preoperatively

Refractive prediction error of AI-driven IOL power model

Mean absolute error (MAE), median absolute error (MedAE), root mean square error (RMSE); proportion of eyes within ±0.5D / ±1.0D refractive prediction error

Time frame: 1 month postoperatively

AI model performance across multi-ethnic cohorts

Stratified refractive prediction error metrics among Chinese, Caucasian and other ethnic subgroups