KM Research Hub

P1 · PANEL / ROI / AXIS GROUND TRUTH

原 EXE 对齐基准真值复核

所有保存均写入追加式审计事件;旧标注、旧比较和旧提取结果不会被删除或覆盖。

返回质量诊断
冻结记录120
真值完整117
待复核0
无效/退役3
← 上一例 Case 83 / 120 · DB #83 下一例 →
artifact_risk invalid_panel_split PMID 42282892

Serplulimab Plus Chemotherapy, with or without HLX04, versus Chemotherapy as First-Line Treatment for Nonsquamous NSCLC: Final Survival Analysis of the Phase III ASTRUM-002 Study.

F2 · Panel A · progression-free survival

Fig. 2. Kaplan–Meier survival curve and subgroup analysis of BICR-assessed progression-free survival, showing the progression-free survival benefits with serplulimab plus chemotherapy compared to chemotherapy alone in the intent-to-treat population and across subgroups. (A) In the ITT population. The median PFS was 12.0 (95% CI 8.7 to 13.9) months in group A, 11.0 (95% CI 8.4 to 12.7) months in group B, and 5.7 (95% CI 4.9 to 6.9) in group C. Serplulimab plus chemotherapy significantly improved PFS compared with chemotherapy alone with an HR of 0.54 (95% CI 0.43 to 0.68; P < 0.001). The addition of HLX04 did not translate to an improvement in PFS compared with serplulimab plus chemotherapy (HR = 1.02, 95% CI 0.80 to 1.28; P = 0.896). (B) Exploratory subgroup analysis of PFS in group B vs. group C, suggesting a PFS benefit with serplulimab plus chemotherapy over chemotherapy alone across the various patient subgroups. (C) Exploratory subgroup analysis of PFS in group A vs. group B, implying a lack of PFS benefit with HLX04 and serplulimab plus chemotherapy over serplulimab plus chemotherapy across the various patient subgroups. Group A, serplulimab plus HLX04 and chemotherapy; Group B, serplulimab plus HLX04 placebo and chemotherapy; Group C, serplulimab placebo plus HLX04 placebo and chemotherapy. Abbreviations: BICR, blinded independent central review; CI, confidence interval; CPS, combined positive score; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; ITT, intent-to-treat; NE, not evaluable; NR, not reached; PD-L1, programmed cell death ligand 1; PFS, progression-free survival; TPS, tumor proportion score.
打开论文记录

坐标框复核

洋红 = 当前 Linux 候选;绿色 = 已保存人工真值。没有绿色框表示仍未形成真值。

Case 83 ROI review overlay
当前候选 {'x0': 60, 'y0': 60, 'x1': 712, 'y1': 682} 人工真值 未保存
生产预期曲线3
当前恢复曲线0
算法版本legacy-0.5.0
坐标 OCR不完整/待核
查看候选 OCR 与警告(只能作种子,不能自证真值)
{
  "complete": false,
  "engine": "RapidOCR",
  "plot_bottom": 751,
  "plot_left": 1,
  "plot_right": 749,
  "plot_top": 364,
  "warnings": [
    "insufficient_x_ticks",
    "insufficient_y_ticks"
  ],
  "x_axis_y": 751,
  "x_confidence": 0.0,
  "x_detected": false,
  "x_max": null,
  "x_min": null,
  "x_ticks": [],
  "y_axis_x": 1,
  "y_confidence": 0.0,
  "y_detected": false,
  "y_max": null,
  "y_min": null,
  "y_ticks": []
}
查看当前 Linux 输出

追加一条审查结论

选择“有效 KM”时,四边 ROI、四个坐标范围、Panel、终点和曲线数均为必填。

像素 ROI(x1/y1 为排他边界)
真实坐标范围

追加式审计历史

最新事件仅作为当前缓存;以下每一条既往结论都永久保留。

事件时间 / 审查人状态ROI坐标说明
#86p1-visual-axis-audit-v2 · frozen benchmark contact sheets + original images + RapidOCR re-audit 2026-07-28T15:12:18+00:00Codex-assisted visual audit invalid_panel_split - - 同一裁剪同时含 KM Panel A 与森林图 Panel B,需拆分后另行纳入。
#3p1-visual-audit-v1 · direct_original_image_review 2026-07-24T17:30:14+00:00Codex P1 visual audit 2026-07-25 invalid_panel_split - - 当前 Panel 资产同时包含上方 A 的三曲线 PFS 图和下方 B 的亚组森林图;必须先重新分割为独立 Panel A,现有整图不得作为单一 KM Panel 真值。图像 SHA-256: 7dbd6c6b5b2116e50d7cba13338c43104268bf7a80261771d9d17ae814d7e97e。

最近一次正式比较

Suite #3 · 2026-07-28T15:15:12+00:00

未通过

真值完整 117 / 120;通过 0;缺真值 0;无效案例 3。