Case Study

Prior vs Posterior — IatrogenicMedications.bayes

Enter patient evidence. Every node shows its prior mean (population baseline) and posterior mean (updated given this patient's values), with the shift highlighted. Nodes held as evidence are marked.

Why prior vs posterior matters

The prior is the population baseline — what the model expects before any patient-specific information arrives. The posterior is the personalised prediction after conditioning on what is actually known about this patient. The shift between them measures the information value of the evidence.

A node that barely updates despite strong evidence signals a missing pathway in the graph. A node that updates sharply even when not directly observed reveals indirect propagation through the network. For clinical decision-making, the posterior is what matters — but the prior provides essential context for judging whether an update is plausible or the patient is genuinely unusual.

Gray bar — Prior mean. The model's population baseline for this node: what we'd expect before knowing anything about the patient.
Teal bar — Posterior mean. The updated expectation after conditioning on the evidence you've entered. For evidence nodes (teal-bordered cards) this is just the value you set. For downstream nodes it's the propagated prediction.
Blue/red shift bar. Magnitude of the update: how far the posterior moved from the prior. Blue = rose, red = fell. The bar length is scaled to the prior value, not the full axis range.
Bar position is shown on a fixed scale covering the node's realistic range (not natural clinical units — these are the model's internal latent indices).
prior mean
posterior mean
shift ↑
shift ↓