The Design And Implementation Of Hadoop Based System For Rational Drug Use

Adverse drug events occur frequently. The relationships of nowadays betweendoctors and patients have become tense increasingly. How to use drugs safely,economically, effectively and rationally, has been a social focus. Because of somemedical personnel are awareness to update their knowledge, weak and even lossconsciousness of responsibility and dedication. Irrational phenomena of drug usestill remains prevalent. Some of them misuse to use antibiotics and other medicines,neglected patients’ physiological and pathological conditions or overlook drugcontraindications, interactions between different drugs. Based on the above, thispaper aims to achieve a rational drug use of the system. When physician’sprescription is not reasonable, this system will give tips in order to change hospitalirrational phenomena of drug use.Firstly, this paper implemented a medical knowledge management system forprocessing text types of data in the original drug directions, finishing forming aprofessional pharmacy data dictionary. The original drug name derives from a totalof200thousands medicare drug directory, and the corresponding instructionscontent is from the official website of China food and drug administration. Theforming process of data dictionary used Chinese word segmentation method. We dida work of word segmentation to the corresponding instruction content based on anexternal Chinese word segmentation dictionary at first, then put the words togetherthrough the manual to form a medicine terminology. The dictionary entry for nowthere are22800and this work took21people per month. Then, this paperimplemented a drug use rationality review system. After receiving a series of datafrom hospital HIS system, this system will compare with the data dictionary afterarranging these data, review the results of unreasonable data to doctor. This systemplays a role of a doctor’s tips. In examining the use of medication, this systemmainly checks from these four aspects, which are patient electronic records, druginteractions, drug contraindications and drug indications. Patient electronic recordsare primarily concerned with patient’s recent drug information, diagnosticinformation. Drug directions are split into pharmaceutical indications,contraindications and drug interactions of drugs three.In addition, the segmentation of medical knowledge management system fordirections content, is combined with mechanical type segmentation method andmechanism of Trie tree dictionary as the theoretical basis, and applies forward maximum matching method. The data dictionary which is organized from theChinese word segmentation tool uses the Hadoop framework to achieve HBasedatabase storage, which is easy to handle large pharmaceutical data. Front-end ofsegmental tools use HTML, JavaScript, Flex and CSS to display, back-end of it useJava to make a unified development. For using comfortably, we choose toencapsulate basic methods of HBase.