The patient has Age: 50.0 ('Age'), has had Pregnancies: 6.0 ('Pregnancies'), has Glucose: 148.0 ('Glucose'), has BloodPressure: 72.0 ('BloodPressure'), has SkinThickness: 35.0 ('SkinThickness'), has Insulin: 125.0 ('Insulin'), has BMI: 33.6 ('BMI'), has DiabetesPedigreeFunction: 0.627 ('DiabetesPedigreeFunction').



You are evaluating a clinical explanation generated by an AI system.

Evaluate the explanation according to these criteria:

1. feature_faithfulness
Does the explanation focus on the important features provided by the explainer?

2. prediction_consistency
Does the explanation support the enhancer prediction?

3. unsupported_claims
Does the explanation introduce unsupported or contradictory medical claims?

4. clarity
Is the explanation understandable for a clinician?

Assign an integer score from 1 (very poor) to 5 (excellent).

Output ONLY valid JSON.

Patient information:
{patient_description}

Enhancer prediction:
{enhancer_prediction}

Important explainer features:
{explainer_ranking}

Generated explanation:
{generated_explanation}

Output format:
{
  "feature_faithfulness": int,
  "prediction_consistency": int,
  "unsupported_claims": int,
  "clarity": int,
  "overall_score": float,
  "short_reason": "..."
}







You are evaluating retrieved medical documents used by a RAG-based clinical explanation system.

Evaluate the following dimensions:

1. retrieval_relevance
Are the retrieved documents relevant to the generated explanation and prediction?

2. grounding
Is the generated explanation supported by the retrieved documents?

3. unsupported_claims
Does the explanation contain claims not supported by the retrieved documents?

Assign an integer score from 1 (very poor) to 5 (excellent).

Output ONLY valid JSON.

Enhancer prediction:
{enhancer_prediction}

Retrieved documents:
{retrieved_documents}

Generated explanation:
{generated_explanation}

Output format:
{
  "retrieval_relevance": int,
  "grounding": int,
  "unsupported_claims": int,
  "overall_score": float,
  "short_reason": "..."
}