Leveraging Pupillometry for Automated Detection of Pediatric Genetic Disorders

Abstract

Pediatric genetic disease diagnostics often involve complex and invasive clinical tests, particularly challenging for infants and young children. This research introduces an innovative paradigm leveraging Chromatic Pupillometry, a technique increasingly recognized for its efficacy in evaluating outer and inner retina functions. The study employs a sophisticated approach, utilizing two distinct Support Vector Machine (SVM) classifiers trained on right and left eye pupil data. Integration of an ensemble voting classifier through OR operations enhances classification accuracy, significantly improving disease detection. Furthermore, the framework is extended to incorporate advanced Long Short Term Memory (LSTM) and Bidirectional LSTM (BI LSTM) algorithms, further elevating the precision and efficiency of pediatric genetic disease identification. This novel methodology not only streamlines the diagnostic process but also reduces the dependence on extensive and invasive clinical tests, offering a promising avenue for early and accurate detection of pediatric genetic diseases.