---
title: CDrugRed
canonical_url: "https://www.modelscope.cn/datasets/xiaojie001/CDrugRed"
md_url: "https://www.modelscope.cn/datasets/xiaojie001/CDrugRed.md"
repository: xiaojie001/CDrugRed
chinese_name: "中文药物推荐数据集"
last_updated: 2026-06-16
license: "Apache License 2.0"
domain:
  - text
tasks:
  - text-generation
language:
  - zh
type:
  - data-to-text
downloads: 96
stars: 0
---

# CDrugRed

> CDrugRed - xiaojie001 在 ModelScope 开源的数据集。CDrugRedtrain.jsonl: 训练集，包含全部字段，供参赛选手用于开发模型 CDrugRedtest-A.jsonl: A榜测试集（即开发集），出院带药列表为空，用于A榜评测测试。注意：A榜测试集答案不公开，不用于模型训练。 候选药物列表.json: 包含数据集全部651种候选药物名 submitpredex.json: 模型预测结果提交格式样例，仅供格式参考，需提交全部预测结果，评测脚本计算指标成功…

xiaojie001/CDrugRed 是 ModelScope 魔搭社区上的text-generation数据集，涉及 text 领域，采用 Apache License 2.0 许可。

- **Repository**: xiaojie001/CDrugRed
- **License**: Apache License 2.0
- **Tasks**: text-generation
- **Domain**: text
- **Downloads**: 96
- **Stars**: 0
- **Last updated**: 2026-06-16

Source: https://www.modelscope.cn/datasets/xiaojie001/CDrugRed

---

# 数据文件说明

- CDrugRed_train.jsonl: 训练集，包含全部字段，供参赛选手用于开发模型
- CDrugRed_test-A.jsonl: A榜测试集（即开发集），出院带药列表为空，用于A榜评测测试。**注意：A榜测试集答案不公开，不用于模型训练。**
- 候选药物列表.json: 包含数据集全部651种候选药物名
- submit_pred_ex.json: 模型预测结果提交格式样例，仅供格式参考，需提交全部预测结果，评测脚本计算指标成功
  
  A榜和B榜均要求提交对应测试集预测结果的Json文件，文件编码为UTF-8，具体格式如下：
  
  ~~~
  [
  {"ID": "1-1", "prediction": ["药物A", "药物B", ...]},
  {"ID": "1-2", "prediction": ["药物C", "药物D", ...]},
  ...
  ]
  ~~~
  
  字段说明：
  
  - ID：对应就诊标识
  - prediction：预测的药物名称列表（须与附件《候选药物列表》一致）







- 详细信息：Chinese Discharge Drug Recommendation in Metabolic Diseases with Large Language Models [!https://arxiv.org/pdf/2510.21084]


@misc{li2025chinesedischargedrugrecommendation,
      title={Chinese Discharge Drug Recommendation in Metabolic Diseases with Large Language Models}, 
      author={Juntao Li and Haobin Yuan and Ling Luo and Yan Jiang and Fan Wang and Ping Zhang and Huiyi Lv and Jian Wang and Yuanyuan Sun and Hongfei Lin},
      year={2025},
      eprint={2510.21084},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2510.21084}, 
}
