We are collecting the COVID-19 Language-based AI research papers and datasets. If you have paper to recommend or any suggestions, please feel free to contact us.

    We released a Chinese medical dialogue dataset (CovidDialog-CH) about COVID-19 and other types of pneumonia. Patients who are concerned that they may be infected by COVID-19 or other pneumonia consult doctors and doctors provide advice. There are 12789 consultations and each consultation has at least 8 senteces.

The whole file is a python list, and each item in the list is a list of dictionary，as shown below:

[{'id': 'dis', 'text': '医生您好！发烧两天，最高38.7， 头痛，全身酸疼乏力，不咳嗽，不流涕，呼吸无异常。现已退烧，头痛基本消失， 但是咽喉及中央气管略有灼烧疼，胸口两侧对称跳着疼，头顶头皮一处跳着疼， 左腹一处偶尔也跳着疼，无力，不爱吃饭。无湖北接触史，偶尔外出去超市采购， 都戴口罩。前段时间压力大，连续一周熬夜做文字工作。请问这是什么情况啊？ 是不是新冠肺炎？谢谢医生！（男，35岁）'}
{'id': 'doc', 'text': '您好！这种情况有多久了？'}
{'id': 'doc', 'text': '有没有鼻塞？'}
{'id': 'doc', 'text': '咳嗽？气喘？'}
{'id': 'dis', 'text': '三天了，前两天发烧头疼，今天不烧了'}
{'id': 'dis', 'text': '只在第一天晚上鼻塞过，不咳嗽，不气喘'}
{'id': 'doc', 'text': '咳嗽？'}
{'id': 'doc', 'text': '咳嗽？'}
{'id': 'doc', 'text': '胸痛？'}
{'id': 'doc', 'text': '应该是病毒性感冒'}
{'id': 'doc', 'text': '不像肺炎'}
{'id': 'doc', 'text': '建议多喝水，休息，按时吃感冒药 如日夜百服宁片或泰诺感冒片，另外可吃些含片，清火的药如清开灵或熊胆胶囊等。'}
{'id': 'dis', 'text': '不咳嗽，胸口两侧对称地方跳着疼，头皮，左腹各有跳着疼'}
{'id': 'dis', 'text': '对了，退烧后，左嘴角上下起了一些疱疹'}
{'id': 'doc', 'text': '还是考虑病毒性感冒'}
{'id': 'dis', 'text': '病毒性感冒是不是传染？家里有孩子，请问怎样防护？'}
{'id': 'doc', 'text': '感冒会传染，'}
{'id': 'doc', 'text': '可以适当隔离，戴口罩'}
{'id': 'dis', 'text': '好的，谢谢医生！'}
{'id': 'doc', 'text': '不客气'}]


The code for read "COVID_Dialogue_Dataset.pk":

with open('COVID_Dialogue_Dataset1.pk','rb') as f:
    data = pickle.load(f)
data[0]

    The DX medical dialogue dataset is built for medical dialogue system, reserving the original self-reports and interaction utterances between doctors and patients.
    We annotate five types of diseases, including allergic rhinitis, upper respiratory infection, pneumonia, children hand-foot-mouth disease, and pediatric diarrhea. We extract the symptoms that appear in self-reports and conversation and normalize them into 41 symptoms. Four annotators with medical background are invited to label the symptoms in both self-reports and raw conversations. Symptoms appearing in self-reports are regarded as explicit symptoms while the others are implicit symptoms. The diseases of each medical diagnosis conversation are labeled automatically.