SPN Learning and Fuzzy-Based Multimodal AI Framework for Ophthalmic Emergency Triage

This dataset accompanies the article "SPN Learning and Fuzzy-Based Multimodal AI Framework for Ophthalmic Emergency Triage" and supports the reproducibility of its experiments. It contains 10,000 patient records [synthetically generated / collected at ...], stored as a single CSV file with 14 variables and no missing values or duplicate records. Each record includes demographic data (age, from 5 to 90 years; sex, coded 0/1), medical history (diabetes and hypertension, binary), and a history of ocular trauma (binary). Seven clinical symptoms are scored on a 0–10 intensity scale: eye pain, redness, blurred vision, photophobia, tearing, discharge, and vision loss. The duration of symptoms is given in days (1–10). The target variable, triage_level, has three urgency classes (0, 1, 2) [where 0 = non-urgent and 2 = urgent; please confirm], with a moderately imbalanced distribution of 44.3%, 35.8%, and 19.9% respectively. The data are intended for developing and benchmarking machine learning, fuzzy logic, and hybrid models for ophthalmic triage and clinical decision support. [The data contain no personal identifiers and are not intended for direct clinical use.]