Mathematical modeling in Newborn Screening
Aims and objectives
In newborn screening diagnostics blood samples taken from newborns within a few days after birth are analyzed for, ideally pre-symptomatic, identification of treatable severe rare diseases to reduce morbidity and mortality. The accurate and efficient diagnosis of these diseases is important and challenging, due to their low prevalence. A quick diagnosis can lead to efficient therapies and treatments which can positively change the outcome and severity of the disease for the children.
The aim of this project is to develop data-driven and theory-driven mathematical models to improve newborn screening for specific inherited metabolic diseases and enhance the understanding of underlying disease mechanisms. Therefore, we closely collaborate with the Division of Pediatric Neurology and Metabolic Medicine at Heidelberg University Hospital to achieve the following goals:
- Data-driven models
- Application of statistical methods to identify pattern within the data
- Development of machine learning algorithms to improve diagnostic accuracy
- Development of Explainable AI algorithms, to clarify the decision making process of the classification model
- Theory-driven models
- Development of metabolic organ-specific whole-body model for infant metabolism
Systematic analysis of newborn screening data and inherited metabolic diseases with metabolic infant model
Research Topics
- Machine Learning and Deep Learning
- Explainable AI
Whole-body metabolic modeling
Collaboration Partners
- EMCL
- Zentrum für Kinder - und Jugendmedizin, Universitätsklinikum Heidelberg
- HIDSS4Health
- Thielelab, University of Galway, Ireland
People from EMCL
- Prof. Dr. Vincent Heuveline
- Elaine Zaunseder
- Saskia Haupt
Contact
Elaine Zaunseder
Publications
- Zaunseder E, Haupt S, Mütze U, Garbade SF, Kölker S, Heuveline V. Opportunities and challenges in machine learning-based newborn screening - A systematic literature review. JIMD Rep. 2022 Mar 23;63(3):250-261
- Zaunseder E, Mütze U, Garbade SF, Haupt S, Feyh P, Hoffmann GF, Heuveline V, Kölker S. Machine learning methods improve specificity in newborn screening for isovaleric aciduria. Metabolites 2023
- Zaunseder E, Mütze U, Garbade SF, Haupt S, Kölker S, Heuveline V. Deep learning and explainable artificial intelligence for improving specificity and detecting metabolic patterns in newborn screening. Accepted SSCI 2023