Developing an AI-Based Model for Lagophthalmos and Bell's Phenomenon Detection in Intensive Care Unit Patients: A Gold Standard Comparison

TL;DR

This study validates an AI model for detecting eye conditions in ICU patients, aiming to prevent corneal damage through early detection.

Contribution

A novel AI model is proposed and validated for detecting lagophthalmos and poor Bell's phenomenon in ICU patients.

Findings

The AI model showed potential for accurately detecting lagophthalmos and poor Bell's phenomenon.

The model could be used by non-ophthalmologists to facilitate early preventive measures in ICU patients.

Early detection may reduce the risk of exposure keratopathy in critically ill patients.

Abstract

Purpose

To validate a newly developed artificial intelligence (AI) model as a screening tool for detecting lagophthalmos and poor Bell's phenomenon, by comparing its performance with the gold-standard examination conducted by an ophthalmologist.

Design

A cross-sectional observational study was conducted. External eye photographs of patients admitted to intensive care units at a secondary hospital in Taif City, Kingdom of Saudi Arabia, were analyzed by the AI model to detect lagophthalmos and assess Bell's phenomenon.

Conclusions and importance

The findings of this study have the potential to significantly impact healthcare by enabling the prediction of serious ocular complications through an AI-based screening model for ICU patients. The integration of AI models can improve patient screening, facilitating early preventive measures such as lid tapes and lubricants to reduce the occurrence of exposure keratopathy. If validated, this AI model could be effectively utilized by non-ophthalmologist staff, including ICU nurses, thereby promoting earlier detection of lagophthalmos and preventing prolonged corneal exposure, which is a known risk factor for exposure keratopathy

Figures