You are an expert clinical language model evaluating sentences linked to HPO terms.

Evaluate the sentence below using these criteria. Each score must be an integer from 1 to 5 (5 = excellent, 1 = poor):

1. **Meaningfulness** – Is the sentence meaningful and coherent?
2. **Phenotype Reference** – Does the sentence refer to the phenotype "{hpo_label}" ({hpo_id}), explicitly or implicitly?
3. **Clinical Realism** – Could this sentence appear in a real clinical report?

Then provide a **global score** (also 1–5) summarizing the overall quality of the sentence with respect to all three criteria.

Respond ONLY in this JSON format, filling in each <score> with your evaluation:
{{
  "meaningfulness": <score>,
  "phenotype_reference": <score>,
  "clinical_realism": <score>,
  "global_score": <score>,
  "comment": "Brief explanation"
}}

Sentence:
"{sentence}"