Fast ophthalmic decision models improve accuracy and speed on key tasks
RetiJev: Fast and Source-Grounded Ophthalmic Decision Models
Computation and LanguageArtificial Intelligence
Summary
Clinical tools that help eye doctors make decisions need to be both fast and accurate while protecting patient privacy. The authors created a new large collection of eye-related questions and developed a set of models called RetiJev that choose the best answers directly, without slow step-by-step text generation. These models answer questions much faster than large language models and perform better on common decision-making tasks. This could help improve emergency or routine eye care by giving doctors quick, reliable support.
What this means in practice
- •For ophthalmic software developers: Integrate RetiJev models to speed up and improve accuracy in automated eye disease diagnosis and guideline grading tasks.
- •For medical device manufacturers: Embed fast decision models to enable real-time clinical decision support in ophthalmic diagnostic devices.$Commercial implications: This enables selling smarter eye diagnostic tools with built-in AI that offers faster and more accurate clinical decisions.
Authors
Bo Liu, Lianyu Wang, Meng Wang, Ching-Yu Cheng, Huazhu Fu
Abstract
Ophthalmic clinical decision support requires fast, accurate, and privacy-preserving inference, yet autoregressive large language models (LLMs) incur substantial latency and computational overhead for decisions among predefined actions. We therefore investigate a decision-oriented alternative to generative reasoning. Specifically, we first introduce RetiJev-Data, a source-grounded ophthalmology corpus comprising 78,158 clinical-decision and diagnosis questions derived from articles in PubMed Central and a MeSH vocabulary covering 531 eye diseases, and 50,074 guideline-grading questions spanning 11 grading systems, along with a 24,662-question evaluation benchmark. Based on this, we develop RetiJev, a family of 0.8B-9B decision models that directly score candidate answers without autoregressive decoding. Across all model sizes, RetiJev outperforms both a 27B LLM and the commercial Jev 1.13 decision model on guideline grading by 10.4-23.0 points. RetiJev-9B further achieves the highest accuracy on clinical decisions, improving over both baselines by 5.4-6.5 points while matching them on diagnosis. It answers a case in 158 ms on one GPU, 3.5x faster than LLM generation and 470x faster than reasoning. Models are released at https://github.com/Awenbocc/RetiJev.