Artificial Intelligence-Driven Telehealth Framework for Detecting Nystagmus.

Area of Science:

• Ophthalmology
• Artificial Intelligence in Medicine
• Medical Diagnostics

Background:

• Nystagmus detection often requires specialized equipment like videonystagmography (VNG).
• Telemedicine offers potential for remote patient care, but requires effective diagnostic tools.
• Integrating AI into clinical workflows can enhance diagnostic capabilities.

Purpose of the Study:

• To implement and evaluate a proof-of-concept AI-driven clinical decision support system for nystagmus detection.
• To assess the system's potential for real-time analysis of clinical data and integration into telemedicine platforms.
• To explore patient benefits including convenience and reduced healthcare costs.

Main Methods:

• A cloud-based deep learning framework was developed to track eye movements and detect facial landmarks in real time.
• The system was trained to analyze data from a bedside clinical test and videonystagmography (VNG).
• Ten subjects participated in this pilot study.

Main Results:

• The AI system's slow-phase velocity (SPV) calculations showed statistical significance (p < 0.05).
• Mean squared error was 0.00459, with a correction error of ±4.8% compared to VNG.
• The system demonstrated accurate real-time eye movement analysis.

Conclusions:

• The deep learning model shows potential for remote diagnostic consultation, possibly supplementing or replacing traditional methods like VNG.
• Advancements in medical AI can improve patient diagnosis, specialist referrals, and physician support.
• Further research with larger sample sizes is warranted for this proof-of-concept study.