You are an expert psychological counseling evaluator. You are given two structured summaries of a psychological counseling case:

- One is the **Ground Truth**, written by a human expert.
- The other is the **Model Output**, generated by an AI model.

Each summary includes the following five attributes:

1. Causes  
2. Symptoms  
3. Treatment process  
4. Characteristics of the illness  
5. Treatment effect

**Ground Truth**:
{ground_truth}

**Model Output**:
{model_output}

Your task is to evaluate each attribute in the model output **independently**, by comparing it to the corresponding section in the ground truth. Use the following four evaluation dimensions:

- **Factual Consistency**: Is the model output factually aligned with the ground truth?
- **Completeness**: Does the model include all key details found in the ground truth?
- **Clarity**: Is the expression clear and coherent?

For each dimension, assign a score from 1 to 5:

- 5 = Excellent
- 4 = Good
- 3 = Fair
- 2 = Poor
- 1 = Very Poor

Then, **write a short comment** (1–3 sentences) explaining your evaluation for that attribute.

Return your evaluation in the following JSON format:

```json
{{
  "causes": {{
    "factual_consistency": X,
    "completeness": X,
    "clarity": X,
    "comment": "..."
  }},
  "symptoms": {{
    "factual_consistency": X,
    "completeness": X,
    "clarity": X,
    "comment": "..."
  }},
  "treatment_process": {{
    "factual_consistency": X,
    "completeness": X,
    "clarity": X,
    "comment": "..."
  }},
  "characteristics_of_illness": {{
    "factual_consistency": X,
    "completeness": X,
    "clarity": X,
    "comment": "..."
  }},
  "treatment_effect": {{
    "factual_consistency": X,
    "completeness": X,
    "clarity": X,
    "comment": "..."
  }}
}}
```