Automatic Detection of Genetic Diseases in Pediatric Age Using Pupillometry

Abstract

Children with inherited retinal disorders have substantial vision impairment. They are divided into outer and inner retina illnesses, and they frequently result in infant blindness. Given the vast range of clinical and genetic symptoms associated with this type of sickness, the diagnosis might be difficult. causes (with more than 200 causal genes). It is frequently based on a complicated network of clinical tests, including invasive ones that are not necessarily suitable for newborns or young children. Thus, a new strategy is required that makes use of chromatic pupillometry, a method that is increasingly being employed to evaluate the inner and outer retinal functions. To assist in the detection of inherited retinal illnesses in pediatric children, this research offers a unique Clinical Decision Support System (CDSS) based on Machine Learning and employing Chromatic Pupillometry. The employment of specialized medical equipment (pupillometer) in conjunction with a specially created custom machine learning decision support system is suggested as a method that blends hardware and software. The features retrieved from the pupillometric data are classified using two separate Support Vector Machines (SVMs), one for each eye. Retinitis pigmentosa in children has been diagnosed using the specified CDSS. The system performed satisfactorily, as evidenced by the results of integrating the two SVMs into an ensemble model, which showed 91.176 accuracy, 1.0 sensitivity, and 0.0 specificity. This study is the first to use pupillometric data to apply machine learning.