Article Preview
TopIntroduction
During the COVID-19 pandemic, online healthcare treatment has been widely used (Yang et al., 2020). Most online consultation platforms are passive service models, requiring patients to choose appropriate departments and doctors according to their conditions. However, it is not easy for most people to go to the appropriate department according to their symptoms. On the one hand, due to the diversity and complexity of medical knowledge, it is difficult for people to make correct decisions due to a large amount of information on the Internet. On the other hand, some people may follow experts blindly, and they may not book the most suitable experts, resulting in a shortage of medical resources. In response to these problems, H. Jiang and Xu (2014) proposed a doctor-recommended method, but this method requires patients to understand their disease, but it is difficult for most patients to describe their diseases accurately. Ju and Zhang (2021) proposed an online pre-diagnosis doctor recommendation model that fused ontology features and disease text mining, but other attributes of doctors and other behavioral information of patients were not considered in this model. To sum up the research, the following problems still exist:
- 1.
Most of the existing studies aim to recommend offline doctors, ignoring the vast potential of online healthcare services.
- 2.
At present, most of the recommendations are based on the objective information of the doctor, ignoring the context of the healthcare services and patients.
In terms of the current form in which the COVID-19 pandemic will coexist with human beings for a long time, online healthcare treatment will continue to flourish, and there will be more and more online healthcare services. As of July 2021, the Haodaifu online platform (https://portal.dxy.cn/), Dingxiang Doctor covers more than 70% of the professional doctor resources in the country. It can be seen that users have requirements, and there are many online healthcare services, but the resources among the platforms are fragmented, and there is no resource exchange. Therefore, it is significant to establish a connection between patients and Internet resources and recommend the most appropriate healthcare service.
This article proposes a multi-dimensional context-aware healthcare service recommendation method, which will recommend the most appropriate healthcare service based on the symptoms described by patients and context factors. Specifically, the contributions of this article are as follows:
- 1.
A multi-dimensional context-aware healthcare service recommendation method is proposed, which will recommend the most appropriate healthcare service for the patient according to the symptoms input by the patient and combining multiple dimensions.
- 2.
A sentiment analysis method for patient reviews is proposed, which uses bidirectional Long Short-Term Memory (BiLSTM).
- 3.
Select multiple indicators from subjective and objective perspectives to construct evaluation models to evaluate doctors and help patients make the most appropriate healthcare service decisions.