The following is a summary of “Inferring skin–brain–skin connections from infodemiology data using dynamic Bayesian networks,” published in the May 2024 issue of Psychiatry by Scutari et al.
Skin diseases and mental illness have been widely studied using cross-sectional epidemiological data. However, this data could only predict associations, not causation, and often focused on a limited set of diseases.
Researchers conducted a retrospective study to complement existing epidemiological evidence by developing a comprehensive causal network model encompassing twelve health conditions.
They used a dynamic Bayesian network to construct the causal network model using Google Search Trends Symptoms data. This model accounted for spatio-temporal trends in the data and could robustly represent cyclic and acyclic causal relationships.
The results showed that the causal network model confirmed numerous cyclic relationships, such as Acne had cyclic relationships with anxiety and ADHD. At the same time, an indirect relationship with depression was also observed. Dermatitis had a direct linkage with anxiety, depression, and sleep disorders while having a cyclic relationship with ADHD. On average, the condition based on the prior week’s data was 0.67. Notable: Acne (0.42), asthma (0.85), ADHD (0.58), burn (0.87), erectile dysfunction (0.76), scars (0.88), alcohol disorders (0.57), anxiety (0.57), depression (0.53), dermatitis (0.74), sleep disorder (0.60), and obesity (0.66).
Investigators concluded that understanding disease interplay and mediators like sleep disorders improves holistic disease management for effective interventions.
Source: nature.com/articles/s41598-024-60937-3#Sec4
The post Using Dynamic Bayesian Networks to Infer Skin–Brain–Skin Connections from Infodemiology Data first appeared on Physician's Weekly.