Nystagmus Detection with an AI-Powered Telemedicine System

This image was created using an AI-powered image generation tool.

Nystagmus is a significant clinical finding in the assessment of vestibular system disorders. A recent study by Harshal Sanghvi and colleagues has demonstrated the feasibility of an AI-powered clinical decision support system capable of detecting nystagmus remotely, in real-time, and at a low cost.

The developed system analyzes video data recorded via mobile devices to evaluate eye movements and guide clinical decision-making processes. The integration of this system into telemedicine platforms holds significant potential as a solution, especially for individuals with limited access to healthcare services.

Limitations of Traditional Diagnostic Methods

Electronystagmography (ENG) and Videonystagmography (VNG), commonly used methods for diagnosing nystagmus, face several limitations. ENG is sensitive to artifacts and may be inadequate in detecting torsional eye movements, while VNG systems are hindered by high costs, large equipment requirements, and patient discomfort, which complicates their widespread use.

AI-based solutions aim to overcome these limitations by offering more accessible, cost-effective, and rapid diagnostic capabilities. When combined with telemedicine technologies, they have the potential to make a significant difference, particularly for communities with limited access to healthcare.

The AI-Powered Telemedicine System and Its Methodology

The AI-powered system presented in the study analyzes data from videos recorded with smartphones using cloud-based deep learning algorithms.

This Python-based system tracks eye movements using 468 facial landmarks with the help of libraries such as MediaPipe's Face Mesh and OpenCV, and it performs slow phase velocity (SPV) measurements. The model has demonstrated highly successful performance with 98% accuracy, a 0.00459 MSE (mean squared error), and a ±4.8% correction error rate. The results are evaluated by a clinician to contribute to the diagnostic and patient counseling processes.

Clinical Application and Future Goals

By combining telemedicine and artificial intelligence, this system has the potential to facilitate diagnosis for patients in remote areas where immediate access to a neuro-ophthalmologist or audiologist is not possible. In the future, the researchers aim to conduct clinical trials on larger and more diverse patient groups to increase the system's validity and to complete the FDA approval process.

Limitations

The study's limitations include the use of a small and homogeneous sample, the impact of environmental factors such as lighting on video quality, and the non-disclosure of the dataset due to patient privacy concerns. Although the deep learning model has shown successful results under controlled conditions, more training data and system improvements are required to enhance its clinical reliability.

References:

1. Sanghvi H, Danesh AA, Moxam J, Reddy SK, Gill GS, Graves BS, Chowdhary S, Chalam K, Gupta S, Pandya AS. Artificial Intelligence-Driven Telehealth Framework for Detecting Nystagmus. Cureus. 2025 May 13;17(5):e84036. doi: 10.7759/cureus.84036. PMID: 40519455; PMCID: PMC12162396.

2. Galoustian, G. (2022, 2 Ağustos). Study first to use eye movements for diagnostic analysis of opioids. FAU News Desk. https://www.fau.edu/newsdesk/articles/eye-movements-diagnostic-analysis