Skip to main content

Applied AI Research

Building systems that solve hard problems — not just publishing papers.


🏥 Healthcare AI📅 2025🎓 Peer-reviewed

Applied Multimodal LLMs for Early Glaucoma Detection

Formal title: Aplicación de grandes modelos de lenguaje en diagnóstico de glaucoma

Vision Language ModelsFine-TuningMultimodal AIOphthalmologyMedical ImagingPython

Designed and fine-tuned a Vision Language Model (VLM) pipeline capable of analyzing retinal fundus images to assist in early glaucoma detection — a leading cause of irreversible blindness worldwide. The system applies large multimodal models to a medical diagnosis task where speed, accuracy, and explainability directly impact patient outcomes.

What was built

A fine-tuned VLM system that ingests ophthalmological retinal images and generates clinically relevant diagnostic outputs. The pipeline integrates multimodal inference, prompt engineering for medical context, and evaluation against expert-annotated datasets.

Technical depth

Fine-tuning large language models on domain-specific medical imagery requires careful dataset curation, compute optimization, and alignment with clinical ground truth. This work addresses each of those constraints in a real healthcare setting.

Why it matters

Glaucoma affects over 80 million people globally. Early detection is critical — yet access to specialist ophthalmologists is limited. AI-assisted diagnosis can bridge this gap at scale, making expert-level screening available earlier and more broadly.

Published at

XLVI Jornadas de Automática (JA-CEA), Cartagena, Spain — 2025

Cite this work

Hernández Vidal, J.; Barrios García, E. J.; Alayón Miranda, S.; Díaz Alemán, V. T. Aplicación de grandes modelos de lenguaje en diagnóstico de glaucoma. Jornadas de Automática, Nº 46, 2025. ISSN-e 3045-4093. https://doi.org/10.17979/ja-cea.2025.46.12085

@inproceedings{vlmglaucomafinetuning2025,
author = {Jorge Hernández Vidal and Eduardo José Barrios García and Silvia Alayón Miranda and Valentín Tinguaro Díaz Alemán},
title = {Aplicación de grandes modelos de lenguaje en diagnóstico de glaucoma},
booktitle = {Proceedings of the XLVI Jornadas de Automática (JA-CEA), Cartagena, Spain, 2025},
year = {2025},
issn = {3045-4093},
doi = {10.17979/ja-cea.2025.46.12085},
language = {Spanish}
}

✨ More applied research on the way.

⬅️ Back to Home