A graph theory approach to analyze birth defect associations

Area of Science:

• Medical Genetics
• Computational Biology
• Public Health Surveillance

Background:

• Birth defects, or congenital anomalies, are prenatal morphological or functional abnormalities.
• Understanding associations among birth defects is crucial for identifying their underlying etiopathogenesis.
• Traditional methods may not fully capture the complex interrelationships between numerous anomalies.

Purpose of the Study:

• To validate the application of graph theory methods for analyzing associations among a comprehensive set of birth defects.
• To explore the utility of graph theory in uncovering patterns and relationships within birth defect data.

Main Methods:

• Utilized birth defect monitoring records from the Latin American Collaborative Study of Congenital Malformations (ECLAMC) spanning 1967-2017.
• Constructed a graph where nodes represent birth defects and edges represent associations, weighted by Volume-adjusted Chi-Square.
• Analyzed graph properties including degree distribution and centrality, and performed graph partitioning.

Main Results:

• The study analyzed 170,430 infants with birth defects from approximately 5 million births.
• The resulting birth defect graph (118 nodes, 550 edges) exhibited a Log-Normal degree distribution, differing from random, scale-free, and small-world models.
• Graph partitioning identified 12 distinct groups, including recognizable syndromes like VATER association, OEIS association, and Patau syndrome. Nonspecific codes like 'Other upper limb anomalies' showed high centrality.

Conclusions:

• Graph theory methods are validated as a robust tool for studying complex birth defect associations.
• This approach can potentially elucidate underlying etiopathogeneses of congenital anomalies.
• The findings suggest graph theory can contribute to improving birth defect classification and coding systems.