Static MONDO candidate queue generated with the same prioritizer logic used by
dismech-mondo-prioritize, with live browser-side reweighting so the score can be inspected and adjusted without regenerating the report.
2 of 7 candidates already have a local dismech root. Remaining candidates still include review and drop recommendations; the bar only reflects current root coverage.
Reweighting changes the score and rank in the browser. Recommended action and specificity bucket stay fixed because they come from the Python heuristics, not the weight sliders. Controls marked as constant or inactive will not change ordering within this export.
These inputs start from conf/mondo_prioritizer.yaml. The exported score remains visible in the table as the baseline for comparison.
The table starts with uncurated rows only. Click any row to inspect the current score contribution list and the fixed heuristics behind its action label.
| Rank | MONDO ID | Label | Current Score | Delta vs Export | Recommended Action | Specificity Bucket |
|---|
Broad parent means child_count >= 4 and the label is not treated as a subtype series or grouping term.
That heuristic can trigger CURATE_ROOT_WITH_SUBTYPES.
Subtype series means the label matches a configured family-stem pattern and that stem appears across at least
2 candidates. If a likely parent is present, the action drops to LUMP_INTO_PARENT.
Over-specific leaf only fires when child_count == 0, the label has at least
7 words, and it contains multiple conjunctions or clause joins.
Missing from dismech and already curated are both kept in the generic scorer so mixed exports can rank curated and uncurated rows together. If one of those factors is constant or inactive in the current export, the weight controls call that out because changing it will not change the ordering.