Machine Learning Techniques for Detecting and Forecasting Disorders in Children Using Pupillometry Data

Abstract

Inherited Retinal Diseases is one among the most significant cause for defects in children resulting in blindness among the children. As diseases of this kind require several clinical test patterns which at most times are inefficient and inappropriate and mostly result in acceptance of alternate methods, as sometime even invasive methods are also used. As this requires a different approach in order to be properly tested and taken care of among the children of the younger age groups and is mostly challenging and difficult to work on. This research deals with a chromatic pupillometry based machine learning approach to satisfactorily diagnose this disease with precise accuracy and specific sensitivity. Several machine learning based decision-making support systems such as Support Vector Machine, BiLTSM, Artificial Neural Network and LTSM are used in this project in order to achieve the predetermined accuracy that is precise and through which sensitive results are obtained specifically based on each eye in particular. As the pupillometric data set is being uploaded into the dataset, it is preprocessed, where a particular prediction model is generated and then sent for filtering. It is in this place where all the unwanted and non-correlative data is separated and only the required data is worked upon.