Abstract

Summary

This study introduces a machine learning-based Clinical Decision Support System using chromatic pupillometry to diagnose inherited retinal diseases like Retinitis Pigmentosa in children, offering a non-invasive alternative to traditional diagnostic methods. For lighting and healthcare professionals, the work highlights pupillometry's growing clinical utility as a non-invasive, objective tool for assessing retinal photoreceptor integrity, including in populations where standard testing is difficult.
Abstract

Key Findings

  • The ensemble SVM model achieved 84.6% accuracy, 93.7% sensitivity, and 78.6% specificity in classifying Retinitis Pigmentosa from chromatic pupillometry data in pediatric subjects.
  • Two separate Support Vector Machines (one per eye) were combined in an ensemble model, with sensitivity notably high, suggesting the system is effective at minimizing false negatives in disease detection.
  • This is reported as the first study to apply machine learning to pupillometric data for diagnosing a genetic disease in pediatric patients.
Categories

Categories

Eye Health & Vision: Paper directly addresses inherited retinal diseases (specifically Retinitis Pigmentosa) in pediatric patients using pupillometry-based diagnostic tools.
The Science of Light: Uses chromatic pupillometry to assess photoreceptor function in inner and outer retina, leveraging spectral light stimuli and pupillary light reflex for diagnostic classification.
Authors

Author(s)

F SIMONELLI, M GHERARDELLI
View more publications