Research Of Recommendation Technology For Massive Healthcare Data

Nowadays, with the explosion of information, Healthcare data is increasing rapidly. In order to get information, one can use search engine and do some searches, but search engine will return much data, which meet the requirement. He/she must scan a lot of documents until he/she find the right one, this contradicts with the demand of quickly find information.In this paper, we propose a healthcare information platform which is used to store, index, search and recommend healthcare information. Healthcare information includes drug information, doctor information, common sense information and prescription. In order to store, index and search information and meet the requirement of high performance & high scalable, we use Elasticsearch, a powerful open source search and analytics engine, and design suitable data structure. As for recommendation, we propose special program for different kind of information. For drug information, we consider three aspects, indication text similarity, usage amount and the popularity of drug usage. For doctor information, we have two strategies, one considers the popularity of doctor, the other considers the similarity of patient praise. For common sense information, we recommend information with the standard of content similarity and user's interest, in order to reduce computation overhead, we should cluster the information. At last, for prescription, user should input symptom descriptions and inspection result, the system will find prescription most approach the inputs, then filter according to user's allergic drug.The experimental results show that the system satisfies the demand of high performance & high scalable. With the introduction of recommendation, response delay is at the same order of magnitude, so real-time performance is perfect.