Bipolar II Disorder — Bayesian Causal Risk Analysis

Multiplicative SCM model · Pearl do-calculus counterfactuals · 26 risk factors · 18 interventions · Sources: PubMed 2003–2024

Longevity Olympics Research Initiative
Severity index
1.00×
Population baseline
Genetic contribution
45%
of risk variance
Somatic contribution
30%
of risk variance
Environmental contribution
25%
of risk variance
Risk factors
Genetic factors Heritability ~58% (BD2)
Somatic / biological
Environmental / psychosocial
Protective interventions
A=pharma RCT/meta · B=psychotherapy/lifestyle RCT. Lower grades excluded from the protective multiplier.
Pharmacological
Non-pharmacological

Model: multiplicative Bayesian risk model using Pearl do-calculus. RR = relative risk vs. population baseline; HR = hazard ratio (protective interventions reduce risk). Gauge uses log₁₀ scale centred at 1.0×. Variance decomposition applies log(RR)-weighting when factors are selected; literature priors (Edvardsen 2008, Lichtenstein 2009) at baseline. Check risk factors present and toggle interventions active, then use counterfactual dropdowns below.

This model is a research/pedagogical tool, not a clinical decision aid.

Severity gauge (log₁₀ scale)
1.00× baseline 0.05× 1.0× 20×
Population baseline
Variance decomposition
Genetic 45% Somatic 30% Environmental 25%
Genetic45%
Somatic / biological30%
Environmental / psychosocial25%
When factors selected: log(RR)-weighted contributions per domain. At baseline: twin-study heritability priors for BD2 (Edvardsen 2008; Lichtenstein 2009). Somatic = independently modifiable biological mediators. Note: independence assumption inflates joint risk for correlated factors — a copula-corrected SCM would attenuate extreme estimates.
Counterfactual analysis (Pearl do-calculus)
What-if scenario — risk factor
Select any factor to see the causal impact of its presence vs. absence:
Select a factor to see its causal impact on severity
If not for — intervention benefit
Select an intervention to quantify its causal contribution to disease control:
Select an intervention to quantify its benefit