Modeling And Optimizing The Cloud System For Regional Appointment Service

Registration difficulty has for so long been an issue in the medical system of our country.Nowadays,it is quite hard for the ordinary people to hang a visit to a specialist of those top municipal hospitals in Shanghai.The most important reason resulting in this phenomenon is the scarcity of medical resources.On the other hand,the imbalance between supply and demand in a very short time,overloads the current medical system.Each municipal hospital in Shanghai has built their own information system for appointment in the past few years.Such kind of service pattern has hindered the share of quality medical resources in Shanghai.And what’s worse is that such mode does not facilitate both the requirements of patients and the healthcare reform.Therefore,in the face of the medical needs of patients in the era of "Internet +",It is very important that how to integrate medical resources and improve the work efficiency of the reservation system so as to contribute to the health of China.With the ongoing development of “Internet + Medical System”,there is an urgent need to raise a new service model for regional medical appointment.We aim at studying the problems on how to integrate all the resources of medical specialists from 38 municipal medical institutions under the Shanghai Hospital Development Center.Constructing a union and convenient service system for appointment is another target of this thesis.Therefore,a comprehensive research is respond to such problem on modeling and optimizing the regional cloud service system to this objective.We mainly discuss the definition,resolution,retrieval and management for disease knowledge,patients’ chief complaints and doctors’ identifications.This thesis is highly relied on the combination of technics among big data analysis,cloud computing and artificial intelligence in the medical domain.We also suggest a new service model to be extended to each regional medical institution.It is aimed at balancing the regional medical resources,improving the clinical experience of patients and achieving the borderless cloud service system for appointment.We study the key problems of modelling and optimizing the cloud service system for appointment.The main methodologies and contributions are listed as followed:1)Studying the methodology for constructing the disease knowledge base deployed in the cloud service system for appointment.We creatively build a new knowledge base,which solves the complexity and diversity of knowledge in the medical domain.Through utilizing the theory and representation of ontology and knowledge graph,we first extract the information from Baidu Index’s Top six open healthcare websites and three large authoritative encyclopedias.Then we optimize the knowledge integration algorithms including designing a new semi-automatic iterative approach.According to the classification categories of disease,we define several new alignment methods for resolving the problems of disease synonyms,type conflicts and data redundancy.Finally,we finish the construction of the disease knowledge base.It is a solid foundation for improving the accuracy of the appointment service.2)Studying the model for parsing chief complaints based on semi-supervised learning.The text of the chief complaint is an important basis for recording the patient’s condition and the patient’s treatment.We creatively design a new semi-supervised learning algorithm to build model for parsing chief complains,which solves the problem of the existing appointment systems.Based on the constructed disease knowledge base,we creatively design eleven possible components of symptoms.These components are further utilized to build symptom syntactic rules and distantly annotate free-text of the chief complains.We also carry out detailed approaches on how to train the Conditional Random Field model and the entity linking algorithm for linking the disease mentions to the existing entities in our knowledge base.Such pipeline accomplishes the structural and normalization of free text and maps all of them to the knowledge base.It can also be used to recognize the symptoms from the chief complaints correctly and recommend a proper department for patients,which eases the whole appointment process.3)Studying the construction of regional doctor resource pool based on the language model and topic model.Doctor resource is the key element of appointment services.We creatively propose a general architecture for building the hierarchy of such kind of resource.We extend the use of IHE-PIX framework to identify doctors and receive the basic information of them.Then,we consider using the extended information of a specific doctor via retrieving the information from the Web.We can now supplement the pool with additional information,providing the patients with much more references of each medical specialist.Finally,we propose to use the Latent Dirichlet allocation(LDA)algorithm to extract short text descriptions of each doctor.These short tags are aimed at carrying out new services for patients to search similar doctors of their own request.Experiments show that the doctor resource pool provide a large quantity of objective information and have a high reference value for patients during their appointment.4)Studying the architecture of the whole cloud system based on PaaS and CDV.We analyze the features of patient resource pool,doctor resource pool and disease resource pool in the appointment cloud service and illustrate the difficulties of sharing these data through the PaaS layer.We give a definite view on the importance of the PaaS layer and describe a preliminary design on how to share these data.Through the utilization of virtual SQL analysis engine and the data grid in CDV,we construct a new model to connect a central node in the regional data center with multiple client nodes in different hospitals.This model can also be deployed quickly in all the hospitals and ease the use of the complex virtual data layer.It can also be used in information exchanging,processing,accessing,computing and management for heterogeneous data including patients,doctors and other information during the appointment service.Through the above four levels of innovation construction,we focus on the modeling and optimizing of cloud system for appointment service,which keeps to the point of appointment on demand.We establish a new service mode for regional appointment service from a global perspective.The theory and methodology proposed in the thesis provides some instructions and experiences in this research area,which have laid foundations for expansion and application in the future.According to a third party scientific search,it is an innovation at home and abroad that the technics and service mode based on 38 municipal medical institutions in Shanghai.This thesis has reached a leading level in this research area.