Management of Patients with Hyperbilirubinemia with Advanced Non-Invasive Monitoring and Artificial Intelligence in the Family-Centered Neonatal Intensive Care Unit

Project no. 23292 · Status: Kapanmış (Yayın bekleniyor) · 2021

Summary

Artificial intelligence is a system that helps solve many problems today. It is an artificial operating system that is expected to exhibit higher cognitive functions or autonomous behaviors specific to human intelligence, such as perception, learning, connecting multiple concepts, reasoning, problem solving, communicating, inferring and decision-making. Within the scope of this project to be developed, it is aimed to classify the data obtained from non-invasive measurements made with a bedside monitor without taking blood from hyperbilirubinemia babies in the Family Centered Neonatal Intensive Care Unit, create value maps and identify risky patients by making predictions, thus contributing to the medical decision-making process.  Conventional methods are used in the diagnosis and monitoring of hyperbilirubinemia, but today, with the developing technology, interest in non-invasive methods is increasing. In this project, software containing artificial intelligence technologies that can make classification and prediction will be developed using evolutionary neural networks, which is a deep learning architecture that is widely used in machine learning model-related issues. Training datasets (VTS) consisting of approximately 45 measurements will be created for each patient using invasive blood tests and non-invasive bedside monitoring measurements. This data set can also be presented for different research as an anonymized open data set specific to Ege University. The Training Data Set will undergo certain pre-processing before being given as input to the model.  Evolutionary neural network model that can classify according to Machine Learning Models; In the next stage, it will be developed to be able to make risk prediction in the diagnosis and follow-up of cases with hyperbilirubinemia, using both open data sets related to hyperbilirubinemia in the literature and data enriched with additional information from the Ege University open data set. The developed system is planned to be integrated into the hospital information automation system to provide easy access to clinical information systems and data analysis applications.

View details