Assessing GPT-4o in Cataract Surgery Decision-Making: 74 Patient Cases

A retrospective clinical study of 74 cataract patients (146 eyes) who underwent surgery, used to evaluate the GPT-4o model's appropriateness for intraocular lens selection. The dataset includes preoperative clinical findings, biometric parameters, systemic histories, and lifestyle questionnaires. It was authored by Yan Liu and last updated on 2026-05-29.

Use Cases

• Benchmarking AI model performance against human surgeons based on appropriateness rates for IOL selection.
• Analyzing the impact of ocular or systemic comorbidities on AI-assisted clinical recommendations.
• Studying the consistency and stability of LLM outputs in repeated clinical decision-making scenarios.
• Training or fine-tuning domain-specific models for routine versus complex cataract surgery planning.

Strengths

• Includes data from 74 real-world patient cases (146 eyes), providing a clinical testbed.
• Contains adjudicated expert consensus labels for model evaluation.
• Results include specific performance metrics (e.g., 37.84% appropriateness rate, Kappa < 0.4, ICC 0.378).

Limitations

• Dataset is very small (19.7 KB), indicating limited scope and sample size.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

figshare, author Yan Liu.

Collection Method

Retrospective clinical observational study; data processed via GPT-4o.

Freshness

Last updated 2026-05-29 06:20:26; freshness should be verified.