Source: NCT04677179_soa.pdf (Protocol J1P-MC-KFAH(b)), 30 PDF pages = document pages 17–46 (PAGEMAP.md). Printed footers run one lower than the document page and were not used. No protocol markdown was available. Prompt 3.8.1, single pass. Four tables extracted.
| # | where | call made | alternative | detail |
|---|---|---|---|---|
| D1 | p17–23, Table 1 row 5 (same in Tables 2–4) | The shaded "Fasting visit" band (X at the fasting visits, reprinted on every page) is a header schedule_property, not an activity | Record "Fasting visit" as a level-0 activity row with X marks | T1.3 |
| D2 | p17, Table 1 Visit number row, marker t1 | Title-cell note "V1 and V9 procedures may be conducted over more than 1 day…" bound to the V1 and V9 header cells | Treat as a table-wide note on the Visit number row | T1.5 |
| D3 | p17, Table 1 rows 6–17 (and top rows of Tables 2–4) | Rows above the first section header (Informed consent … Alcohol and caffeine use; Concomitant medications … Tobacco use in Tables 2–4) recorded at indentation level 0 as stand-alone activities | Level 1 under an implicit unnamed group | T1.4 |
| D4 | p24, Table 2 | Typed track, track_label "Responders" (schedule applies only to responders at Week 12; exclusive with Table 3) |
main_soa as an independent maintenance schedule |
T2.1 |
| D5 | p28, Table 2 (continued) | "Table 2 (continued)" (V20–V29, p28–32) treated as the second column tile of one tiled Table 2; rows unioned by name, marks merged; redacted "CCI" rows matched across tiles by position between the same neighbours | Separate continuation table, or a different matching of the redacted rows | T2.2 |
| D6 | p24/p28, Table 2 Visit number row, markers t5/t6 (and t2/t3) | Remote-visit note bound to the visit cells it names (V10–V19 tile 1 as t5; V20–V28 tile 2 as t6, V29 excluded). Tile variants that differ in wording/punctuation kept as separate notes (remote note: one comma; ETV cross-reference: "an ETV" vs "an early termination visit") | Merge the variant pairs, and/or bind to the whole Visit number row | T2.5 |
| D7 | p27, Table 2 "Endoscopic Procedure" header, marker c30 | Comment "Not applicable for responders during the time period of V10 through V19" bound to the section-header row (governs the group) | Bind to the Endoscopy and Colon biopsy rows, or to the V10–V19 columns | T2.5 |
| D8 | p33, Table 3 | Typed track, track_label "Nonresponders" (applies only to nonresponders at Week 12) despite identical V10–V29 numbering to Table 2 |
domain of Table 2 or main_soa |
T3.1 |
| D9 | p38, Table 3 (continued) | "Table 3 (continued)" (V20–V29, p38–42) treated as tile 2 of one tiled Table 3; rows unioned; "CCI" rows matched by position | Separate continuation table / different matching | T3.2 |
| D10 | p33/p38, Table 3 Visit number row, markers t5/t6/t7 | "Visits 17 through 28…" note bound to V17–V19 (t5) and V20–V28 (t6, tile-2 text lacks final period); "V16 procedures…" note bound to the V16 cell (t7) | Table-wide notes on the Visit number row; merge the two remote-visit variants | T3.5 |
| D11 | p43, Table 4 | Typed main_soa (independent schedule, own columns ETV/V997/V801/V802, applies to all participants) |
track (e.g. "Follow-Up") as a separate study-phase timeline |
T4.1 |
| D12 | p43, Table 4 Visit number row, markers t1/t2 | "V802 is only for randomized participants who were positive for anti-HBc…" bound to the V802 cell; "V997: Additional study procedures…" bound to the V997 cell | Table-wide notes on the Visit number row | T4.5 |
| D13 | p43, Table 4 Vital signs, marker c4 | Vital-signs comment (ending "V997: Vital signs collection is optional.") kept as one footnote on the row | Split the V997 sentence into a note on the Vital signs × V997 cell | T4.5 |
footnote (or source_note for the bare "See Section 8.2.2."), deduplicated by text within each table, with synthesised markers c1… (activity_name / schedule_property locations, method synthesized).t1…; pure cross-references ("For procedures at an ETV, see ETV in Table 4." etc.) typed source_note; population statements and cross-references with no single visit bound to the Visit number schedule_property (method synthesized).glyph_reconstruction / deglyph_reconstruction.continuation tables (the call itself is open as D5/D9).pdftotext -bbox, X tokens (x > 250 pt, excluding the Fasting-visit band) binned to header column centres. Totals: T1 158, T2 180, T3 211, T4 54, equal to the activity_schedule counts. Lowercase "x" tokens in the bbox stream were fragments of words (e.g. "exam") in label/comment columns and were excluded by position.font_signal); rows under them = level 1; stand-alone rows above the first header = level 0 (visual_estimate, D3).main_soa. 9 data columns (V1–V9, positions 2–10). 5 schedule properties, 63 activity rows (8 section headers), 158 marks, 32 annotations.text_match; D2), t2/t3 source notes on the Visit number row; c1–c29 Comment-column notes. c23 (PK collection timing) binds two rows (PK samples and the following CCI row) — two separate cells with identical text, deduplicated.track, track_label "Responders" (D4). 20 data columns (V10–V29, positions 2–21). 40 activity rows, 180 marks, 19 annotations.track, track_label "Nonresponders" (D8). 20 data columns (V10–V29, positions 2–21). 40 activity rows, 211 marks, 18 annotations. No epoch row separates Extension Induction from Extension Maintenance, so none is recorded.main_soa (D11). 4 data columns (ETV, V997, V801, V802; positions 2–5). 39 activity rows, 54 marks, 16 annotations.activity_name_source.method: glyph_reconstruction) and Comment-column notes (annotation_text_source.method: deglyph_reconstruction) where every occurrence of the note is on a raster page. Where a note also prints on a vector page, text was taken from the vector page and no method is recorded. Words were re-segmented and checked against the rendered pages and against the same wording on the vector pages (e.g. "peripheral lymph nodes", which the text layer runs together as "peripherally mph").annotation_markers. No marker referenced without definition (the tables print no superscript markers at all; all markers are synthesised).activity_name_source.method: glyph_reconstruction — every activity whose source_page is a raster page (doc 18, 21, 22, 23, 24, 26, 27, 28, 30, 31, 33, 35, 36, 37, 38, 40, 41, 43, 44, 45).activity_name_source.indentation_method: font_signal for headers and grouped rows; visual_estimate for the ungrouped level-0 top rows (D3).annotation_text_source.method: deglyph_reconstruction — raster-only Comment notes (listed per annotation in the JSON).marker_locations.method: synthesized — all Comment-column and table-wide title notes; text_match — title notes bound to the visit cells they name (T1 t1; T2 t5, t6; T3 t5, t6, t7; T4 t1, t2).proximity, no proximity_bounded, no unresolved locations. No schedule_property structure_method recorded (all header rows carry printed labels).