Nephrotoxicity Explorer: test evidence Experimental

NEP-ZONE-001 The stage zones are painted as filled regions from the staging cut-points: a Stage-3 band above the last fold cut, a Stage-2 band between the second and third, and Stage 1 as the band between the first and second PLUS everything at or above the absolute trigger below the first — the L. The no-stage box is left unpainted. Regions are clipped to the visible domain and omitted when it excludes them.

NEP-ZONE-002 Zones are drawn in beforeDatasetsDraw, worst stage first, so every participant is drawn on top of the region that describes them.

NEP-ZONE-004 A control hides the zone labels without removing the zones, for a reviewer reading the cloud rather than the regions.

NEP-STAGE-001 The fold-change ladder stages a participant at 1.5× (Stage 1), 2× (Stage 2) and 3× (Stage 3), read WORST-MATCH-FIRST so no cut-point can be shadowed by a lower one.

NEP-STAGE-004 The combined stage the zones show is the WORSE of the two axes, raised to Stage 3 by the absolute-value rule; a suppressed absolute-change stage cannot lower the fold stage.

NEP-UNIT-002 Results convert to the target unit PER RECORD (1 mg/dL = 88.4 µmol/L) before any comparison, so one participant's records may mix known units and still stage correctly.

NEP-SCAT-001 One mark per participant, filled by the combined stage so the cloud reads by severity even with the zones hidden, with a distinct larger marker for a participant tripping the absolute-value rule.

NEP-STAGE-003 A participant whose maximum creatinine reaches ≥ 4.0 mg/dL is Stage 3 whatever their fold change says. The rule is a property of the participant — the value reached, not a change — so it is carried on the mark and in the tooltip rather than as a region of the plane, and is counted as Stage 3 in the summary table.

NEP-SCAT-004 Clicking a point selects it, styles it distinctly, names the participant and their stage on the chart, and dispatches the shared participantsSelected event on the shell root; clicking empty canvas with a selection active clears it and dispatches the empty selection.

NEP-TBL-001 The summary table gives Stage 0–3 down the side and N + % across for the fold-change staging, the absolute-change staging, and the combined stage the zones show.

NEP-TBL-002 Every stage row renders even when empty, and the absolute-change column's Stage 2 and Stage 3 cells are marked as not applicable rather than zero, with a note saying KDIGO defines no such stages on absolute change.

NEP-DATA-005 Records and participants that cannot be plotted are counted AND exportable, never silently dropped: a missing or non-numeric result leaves as a row with its reason attached, a participant with no usable result or no post-baseline record leaves as a participant, and each count carries a CSV download built at click time. The module's own derived working is excluded from the export.

NEP-CFG-006 Field-list settings (filters, details) accept column names or { value_col, label } specs; the detail list defaults to the participant id plus the filter columns; and the rendered filter controls constrain the plotted population on the active selections only, restating the summary table.

NEP-UNIT-003 When any record's unit is absent or unrecognized the module refuses to guess: the fold-change axis stays (it is a ratio and unit-free), and the absolute-change staging, the absolute cut-line and the ≥ 4.0 mg/dL rule are all suppressed, with the y-axis labelled in the native unit and a note above the chart saying so. Within that mode a participant whose own records use more than one unit has no computable fold change and is dropped with a reason.

NEP-CFG-001 The default settings carry the long-format lab mapping the KDIGO scatter needs: a participant id column giving one point per participant, plus the measure, value, unit, visit and numeric-visit columns.

NEP-CFG-002 Baseline identification is an explicit baseline_col / baseline_value pair, defaulting to no configured flag.

NEP-CFG-003 The stage cut-points are a stages setting — three fold-change cut-points, one absolute-change trigger, one absolute-value rule — defaulting to the KDIGO criteria (1.5 / 2 / 3, 0.3 mg/dL, 4.0 mg/dL). Supplied members merge onto the defaults, and custom cut-points restage the same data.

NEP-CFG-004 The fold ladder is numified and sorted ascending on merge, and a ladder that is not three usable numbers falls back to the KDIGO default rather than half-applying.

NEP-CFG-005 The unit contract is a units setting with a target unit (mg/dL) and a per-unit factor table; caller-supplied factor keys are normalized on merge so a lookup is a plain map read.

NEP-CFG-006 Field-list settings (filters, details) accept column names or { value_col, label } specs; the detail list defaults to the participant id plus the filter columns; and the rendered filter controls constrain the plotted population on the active selections only, restating the summary table.

NEP-CFG-007 The stage-zone labels are a shown / hidden setting defaulting to shown, and an unrecognized value falls back to shown rather than silently hiding them.

NEP-CFG-008 The creatinine measure resolves through measure_values.CREAT, so a dataset that names the measure differently is configured rather than patched, and the creatinine check follows the renamed measure.

NEP-UNIT-001 Unit strings are normalized before matching — trimmed, lower-cased, with all three micro spellings (u, µ U+00B5, μ U+03BC) folded — so one factor-table entry covers every spelling and casing.

NEP-STAGE-001 The fold-change ladder stages a participant at 1.5× (Stage 1), 2× (Stage 2) and 3× (Stage 3), read WORST-MATCH-FIRST so no cut-point can be shadowed by a lower one.

NEP-STAGE-002 The absolute-change axis carries exactly one cut-point (0.3 mg/dL) and produces Stage 1 only; there is no Stage 2 or Stage 3 on absolute change.

NEP-STAGE-003 A participant whose maximum creatinine reaches ≥ 4.0 mg/dL is Stage 3 whatever their fold change says. The rule is a property of the participant — the value reached, not a change — so it is carried on the mark and in the tooltip rather than as a region of the plane, and is counted as Stage 3 in the summary table.

NEP-STAGE-004 The combined stage the zones show is the WORSE of the two axes, raised to Stage 3 by the absolute-value rule; a suppressed absolute-change stage cannot lower the fold stage.

NEP-DATA-001 When a baseline-flag column is configured, the participant's first record whose value equals baseline_value is the baseline, outranking record order.

NEP-DATA-002 With no flag configured — or for a participant whose flag matches no record — the earliest record is the baseline: study day first, then visit number, then input order. The number of participants resolved by that fallback is reported rather than assumed.

NEP-DATA-003 Each participant reduces to one point: their baseline, their maximum value over the POST-baseline records (the baseline record itself can never be the maximum), and the fold change and absolute change between them.

NEP-DATA-004 A participant whose creatinine only ever fell keeps a negative absolute change and a fold change below 1, and is plotted rather than dropped.

NEP-DATA-005 Records and participants that cannot be plotted are counted AND exportable, never silently dropped: a missing or non-numeric result leaves as a row with its reason attached, a participant with no usable result or no post-baseline record leaves as a participant, and each count carries a CSV download built at click time. The module's own derived working is excluded from the export.

NEP-UNIT-002 Results convert to the target unit PER RECORD (1 mg/dL = 88.4 µmol/L) before any comparison, so one participant's records may mix known units and still stage correctly.

NEP-UNIT-003 When any record's unit is absent or unrecognized the module refuses to guess: the fold-change axis stays (it is a ratio and unit-free), and the absolute-change staging, the absolute cut-line and the ≥ 4.0 mg/dL rule are all suppressed, with the y-axis labelled in the native unit and a note above the chart saying so. Within that mode a participant whose own records use more than one unit has no computable fold change and is dropped with a reason.

NEP-TBL-001 The summary table gives Stage 0–3 down the side and N + % across for the fold-change staging, the absolute-change staging, and the combined stage the zones show.

NEP-TBL-002 Every stage row renders even when empty, and the absolute-change column's Stage 2 and Stage 3 cells are marked as not applicable rather than zero, with a note saying KDIGO defines no such stages on absolute change.

NEP-TBL-003 An empty population summarizes to zeroes rather than dividing by zero.

NEP-SCAT-003 The point carries the visit and the study day its maximum came from; the study-day line is omitted rather than rendered blank when the data has no study-day column.

NEP-ZONE-005 The axes keep every cut-point on screen: the fold axis runs from 0 to past the last fold cut-point, the change axis covers zero and the absolute trigger and extends below zero when the data goes there, both are padded so no point sits on the frame, and the cut-points are labelled as ticks.

NEP-ZONE-001 The stage zones are painted as filled regions from the staging cut-points: a Stage-3 band above the last fold cut, a Stage-2 band between the second and third, and Stage 1 as the band between the first and second PLUS everything at or above the absolute trigger below the first — the L. The no-stage box is left unpainted. Regions are clipped to the visible domain and omitted when it excludes them.

NEP-ZONE-002 Zones are drawn in beforeDatasetsDraw, worst stage first, so every participant is drawn on top of the region that describes them.

NEP-ZONE-003 Stage severity is ordinal, so the zones take an ordered status ramp at a background-appropriate opacity, and every zone carries a text label — colour is never the only cue.

NEP-SCAT-001 One mark per participant, filled by the combined stage so the cloud reads by severity even with the zones hidden, with a distinct larger marker for a participant tripping the absolute-value rule.

NEP-SCAT-002 The point tooltip carries the source's lines: participant, KDIGO stage, fold change with its stage, absolute change with its stage, baseline and maximum creatinine with the visits they came from — signing a negative change with a real minus glyph — plus a line naming the absolute-value rule when it fires.

NEP-SCAT-004 Clicking a point selects it, styles it distinctly, names the participant and their stage on the chart, and dispatches the shared participantsSelected event on the shell root; clicking empty canvas with a selection active clears it and dispatches the empty selection.

NEP-DATA-006 checkInputs validates the schema's required column settings (id, measure, value) against the remapped names and throws naming every missing column; a dataset carrying no creatinine at all is reported in the chart's annotation rather than thrown.

NEP-API-001 The public module collection exposes the nepExplorer factory, and the module ships the standard lifecycle (init / setData / setSettings / render / resize / destroy) over the shared shell.

NEP-COHORT-001 The synthetic acute-kidney-injury cohort is present in the shared demo extract, labelled synthetic by participant id, site and arm, and carries the file's measure contract with creatinine as its only measure.

NEP-COHORT-002 Every cohort participant has a day-0 baseline record and enough distinct post-baseline visits to trace a path.

NEP-COHORT-003 Cohort creatinine is written in µmol/L, in the single unit spelling the rest of the file uses for that measure.

NEP-COHORT-004 The demo extract carries no baseline-flag column, so the demo runs on the baseline fallback.

NEP-COHORT-005 All four fold-change stages are populated, with enough participants per stage that the summary table's percentages mean something.

NEP-COHORT-006 The cohort's absolute-change staging carries Stage 1 and no-stage participants only.

NEP-COHORT-007 The fold and absolute stagings disagree in BOTH directions: some participants are Stage 1 on absolute change alone, and some on fold change alone.

NEP-COHORT-008 Every zone the chart paints is populated on the combined stage.

NEP-COHORT-009 The cohort contains participants who reach ≥ 4.0 mg/dL, including participants whose fold change alone would only be Stage 1.

NEP-COHORT-010 The cohort contains participants whose creatinine only falls, whose maximum post-baseline value is clearly below their own baseline.

NEP-COHORT-011 Both arms are populated and the injury signal skews to the active arm.

NEP-COHORT-012 The injection ADDS rows and leaves the pharmaverseadam and chronic-liver-disease populations byte-for-byte untouched; the generator is deterministic and idempotent, and fails the build when a generated participant misses the stage its archetype promises.