Plan quality.

"Is this plan good?" is a question about a population, not one plan. Radial computes plan-quality metrics through curation and reads them across the whole cohort as distributions, so an outlier is visible against the cases around it, not judged in isolation.

Two kinds of quality signal

Delivery-complexity metrics are computed as a case imports. The reference-based scores that need the resolved structure set (e.g., conformity, homogeneity, and target coverage) are computed once a case is curated, so they rest on confirmed names rather than raw labels.

Quality as a cohort distribution

Plan-quality metrics live in the cohort views, not in a single plan's readout. Radial renders each one as a distribution across the cohort, so the question becomes concrete: here is where this plan sits among its peers, and here is whether the cohort as a whole is drifting.

Reading a distribution honestly

A distribution says more than a single pass/fail: a case-mix difference reads differently from a systematic offset. Either way the individual cases are one click away, because every point in every distribution is a patient you can open.