Sample 128

Evaluation Instructions

Task Overview: AI models read clinical trial protocol documents and were asked to extract evidence supporting specific eligibility criteria modifications. You're evaluating whether the AI extracted the RIGHT evidence that justifies the post-modified (final) eligibility criteria.

Workflow: AI reads protocol → AI extracts evidence → Ground truth is final modified criteria → You evaluate if AI's evidence actually supports those final criteria

Important: Some outputs may be low quality and should be scored accordingly.

Your task: Compare the model-generated prediction (right panel) against the ground truth criteria (left panel).

Clean view: Shows formatted, readable text for easy comparison
Diff view: Highlights differences between ground truth and prediction
Raw view: Shows original unformatted text as it appears in the dataset

Evaluation scale (0-4):

0 = No match: Prediction is unrelated or contradicts the target criteria
1 = Poor match: Minimal overlap with ground truth, major errors or omissions
2 = Partial match: Some key elements captured, but notable gaps or inaccuracies
3 = Good match: Most elements correct, minor differences or formatting issues
4 = Excellent match: Essentially identical content, fully correct capture

Consider both content accuracy and completeness. Some predictions may be technically "correct" but incomplete or out of order.

Ground Truth

INCLUSION CRITERIA

  1. Eastern Cooperative Oncology Group (ECOG) performance status (PS) 0 or 1.
  2. Participants must have a pathologic diagnosis of classical Hodgkin lymphoma (cHL) who are relapsed or refractory with one of the following:.i) Autologous stem cell transplant (ASCT) ineligible patients.ii) Patients after failure of ASCT.- Must have at least one lesion that is > 15 mm (1.5 cm) in the longest diameter and avid by Fluoro Deoxy Glucose (FDG) Positron Emission Tomography (PET) scan.

EXCLUSION CRITERIA

  1. Known central nervous system lymphoma.
  2. Participants with nodular lymphocyte-predominant Hodgkin lymphoma (HL).
  3. Participants with known history of pancreatitis or progressive multifocal leukoencephalopathy (PML).
  4. Other protocol-defined Inclusion/

EXCLUSION CRITERIA

  1. apply.

Prediction

INCLUSION CRITERIA

  1. Eastern Cooperative Oncology Group (ECOG) performance status (PS) 0 or 1.
  2. Participants must have a pathologic diagnosis of classical Hodgkin lymphoma (cHL) who are relapsed or refractory with one of the following:
  3. i) Autologous stem cell transplant (ASCT) ineligible patients.
  4. ii) Patients after failure of ASCT.
  5. Must have at least one lesion that is > 15 mm (1.5 cm) in the longest diameter and avid by Fluoro Deoxy Glucose (FDG) Positron Emission Tomography (PET) scan.
  6. Participants must not have received prior treatment with an anti-PD-1, anti-PD-L1, anti-PD-L2, anti-CD137, or anti-CTLA-4 antibody.

EXCLUSION CRITERIA

  1. Known central nervous system lymphoma.
  2. Participants with nodular lymphocyte-predominant Hodgkin lymphoma (HL).
  3. Participants with known history of pancreatitis or progressive multifocal leukoencephalopathy (PML).
  4. Participants with evidence of active brain metastases on brain imaging (i.e., MRI or contrast CT).
  5. Participants who have received prior allogeneic transplant.
  6. Participants who are BV refractory as identified by a failure to achieve a PR or CR, relapse, or progression within 3 months of receiving BV treatment.
  7. Participants with any positive test result for hepatitis B virus or hepatitis C virus indicating presence of virus.
  8. Participants with known history of testing positive for human immunodeficiency virus (HIV) or known acquired immunodeficiency syndrome (AIDS).
  9. Participants who are compulsorily detained for treatment of either a psychiatric or physical (e.g., infectious disease) illness.
  10. Females who are pregnant or breastfeeding.