Systemic Amyloidosis — All Types, Subtypes & Variants

Eigenvalue diagnostics for the selected interventions under an equicorrelation matrix (off-diagonal ρ̄). A large λ_max relative to k signals redundancy; n_eff is the effective number of independent interventions actually contributing.

The cross-correlation is now mostly STRUCTURAL, not a single number. Interventions on the same cascade node (e.g. three TTR silencers at node 1) are near-perfect substitutes and saturate against each other within that node; interventions on different nodes (a silencer at node 1 + a stabiliser at node 2) are d-separated given the intermediate node and compose in series, so they stack — which is exactly why HELIOS-B saw real incremental benefit adding vutrisiran on a tafamidis background. The residual ρ̄ slider only governs the leftover correlation among the mediator-independent (direct) supportive effects.

Minimum-effective-set analysis. Set a target combined risk reduction; the model finds the smallest set of interventions — accounting for front-door mediator overlap — that reaches it, and highlights them. If the target exceeds what all interventions together can achieve, the full set is shown (never an empty one). "Apply" ticks exactly that set.

Monte Carlo propagation. Each selected intervention's log-effect is sampled from a normal distribution implied by its 95% CI; samples are combined with the same eigenvalue overlap discount. 5,000 draws.

Antithesis — challenging this oracle's own conclusions

What-if — the do-operator: P(Y | do(S))

Intervening on the selected set S with Pearl's do-operator (setting the interventions, not merely observing them). Contrast against do(∅) = no intervention.

If-not-for — but-for counterfactual (leave-one-out)

For each intervention: "if not for this one, the combined front-door effect would be…". Isolates each intervention's marginal causal contribution after mediator-overlap removal, so shared-pathway agents are not double-credited.

One-at-a-time sensitivity. Each intervention's effect is swung across its 95% confidence interval (others held at point estimate); the bar is the resulting swing in the combined front-door effect. A long bar means the combined estimate leans heavily on that single trial's precision.

Front-door (mediation) decomposition over the explicit cascade. Each intervention’s mediated log-effect is pooled WITHIN its node with dose-response saturation (same-node substitutes), then the nodes are composed in SERIES down the flux (production → stability → fibril → burden → organ), with a series attenuation because reducing an already-reduced flux buys less. This replaces the old single-mediator global saturation: a silencer + a stabiliser no longer wrongly cancel as if competing for one pool. Direct (organ-level) effects keep the residual eigenvalue ρ̄. The per-node reductions are reported so the channel structure is visible.

Which % of cross-correlation is appropriate? Not one number. The mediator overlap is fixed empirically by the Amyloid saturation (currently removing — of the summed mediated effect when interventions are stacked). Note the failure modes on display: birtamimab moved a surrogate and a sub-group but its confirmatory trial failed; SGLT2i and the transplant arms move the outcome on observational, selection-confounded data. Neither should be read the way an RCT ACM result (vutrisiran, tafamidis, daratumumab) is read.

Front-door caveat (antithesis): the mediator→outcome edge is confounded by stage and organ pattern; effect sizes are stage-conditional; and the whole graph is valid only WITHIN one correctly-typed amyloidosis. The all-types view is a landscape for comparison, not a poolable model — a combined figure spanning types is shown for reference only and flagged.