Speaker
Description
The DIPPS project endeavors to construct a digital platform aimed at supporting psychological and psychiatric interventions to improve mental health. The DIPPS platform encompasses various functionalities, from the digitalized assessment data to a predictive system for potential symptom exacerbation.
Utilizing the digitalized assessment data, Bayesian network (BNs) can be informed to forecast the pathways and probabilities leading to worsening of specific psychopathology indicators. BNs represent a novel method to modeling causal relations, especially with cross-sectional data, where partial correlation networks are challenged. Employing a directed acyclic graph (DAG), BNs depict causality and conditional probabilities.
In a sample of 1017 participants, the hill climbing (HC) algorithm was used with the R-package “bnlearn”. Initially, the HC algorithm starts with a saturated network and progressively introduces tests that augment the number of conditioning nodes. To ensure a robust BN, a bootstrap was employed with 1000 iterations. The final averaged network incorporated edges that recurred in at least 85% of the bootstrap iterations and maintained consistent directions in over 50% of these iterations. Findings indicated that intolerance of uncertainty is a ‘parent node’, influencing anxious, somatic and PTSD symptoms, while feelings of loneliness are a 'child node,' influenced by most other nodes.
Bayesian networks facilitate researchers in identifying plausible causal connections within psychometric data, providing fresh insights into the psychological phenomena under examination. They are instrumental in formulating novel hypotheses and substantiating existing ones with relevant evidence. Insights gained from BNs could inform more effective interventions or personalized treatment strategies for mental health conditions.
| If you're submitting a poster, would you be interested in giving a blitz talk? | Yes |
|---|
