Mapping psychiatric comorbidity network: A pilot multi-method weighted network analysis with a focus on key disorders

Area of Science:

• Psychiatry
• Network Science
• Computational Health

Background:

• Psychiatric comorbidity presents a significant burden, yet research often focuses on limited associations.
• Understanding complex comorbidity patterns is crucial for effective healthcare.
• This study addresses limitations by employing network analysis for a comprehensive view.

Purpose of the Study:

• To construct a psychiatric comorbidity network using network analysis.
• To explore network structural characteristics under varying weight definitions.
• To identify central disease categories and their interrelationships.

Main Methods:

• Utilized psychiatric outpatient data (Jan 2016-Jun 2024) from Changhua Christian Hospital, Taiwan.
• Extracted ICD-10 diagnostic codes (F00-F99) for patients with diagnoses appearing at least three times.
• Constructed three networks (co-occurrence, Jaccard index, mixed graphical model) and analyzed centrality and community structure.

Main Results:

• Analysis of 16,954 patients revealed distinct network structures based on weighting methods.
• Mood (F34) and anxiety (F41) disorders showed high centrality in the co-occurrence network.
• Developmental disorders (F8x) gained centrality in the Jaccard index network, while substance use disorders (F1x) were central in the mixed graphical model network.

Conclusions:

• Different network weighting methods highlight the varying roles of psychiatric disorders in comorbidity.
• Findings offer novel perspectives on disease interrelationships and support clinical practice.
• The study provides valuable data for future research on psychiatric comorbidity networks.