Abstract
Physicians make diagnosis as well as treatment decisions. This study is aimed at employing systems that help them make these decisions. To achieve this, the study utilizes web services in combination with Bayesian theorem as well as decision trees to put up a web-services-based system, which helps in decision support. Physicians key in available probabilities and obtain results (diagnosis) through computational analysis that employ Bayesian’s theorem. This can then be processed in the form of XML by decision-tree-based support service. The outcome is a recommendation along with analysis for quality diagnosis. The platform offered by this system is more of a service than conventional; it can be accessed through any operating system that can access the internet via HTTP. It will also benefit everyone as its use will be free on the internet, once completed. In a way, this achieves the main objective of the study, to provide and integrate heterogeneous medical support systems (Chang and Lu 415-422).
Introduction
This study is aimed at employing web services to provide a platform for diagnosis as well as recommendations for treatment. The ultimate objective is to provide freely, this service to the population so as to speed up decision support service processes. The study combines two theoretical frameworks, web service and medical decision support. The framework is founded on Decision trees as well as Bayesian theorem; it is expected to provide concrete solutions to medical decision problems freely for everyone. Technology employed is web services technology with its functions constructed using MS.NET. XML and web services provide communication capacity under which the web technology allows XML to interact with other platforms for its usability. This also allows programs with simple services to transmit complex value-added services. This study therefore aims to create a medical decision support system that is based on web services to improve medical decision support for efficiency and expedition. This study will try to build a medical decision support service system, which is based on web platform using decision trees and Bayesian theorem (Chang and Lu 415-422).
Stages of Medical decision support systems
This study employs web services to improve medical decision support service for all. The components involved include web service technology, XML, Bayesian theorem as well as decision tree support. This is mainly because web services allow cross –platform applications. When this is achieved, then other users accessing the web can utilize the service. The system uses Bayesian theorem for its analysis through XML message transmission. The outcome (from the two processes) is then transmitted to treatment decision support for analysis as well as for the users to make proper diagnosis and treatment decisions. This has the essence of improving quality of medical decisions as well as enhancing efficiency. In addition, the speeds at which these processes are conducted are improved for better quality medical services. This framework is therefore necessary for implementation to improve service delivery for medical decision support systems (Chang and Lu 415-422).
The method employed in this study is empirical. This allows sharing of services through internet, the framework (medical decision support service) that utilizes this service (internet) is created and its feasibility verified through the internet. Moreover, medical decisions support services, SOA along with web services are organized based on theoretic and framework explored in existing literatures. This study involves several processes, and these include exploration and analysis of associated theories, investigating ways that would assist in integrating the theories and framework that are related to the existing tools which can be employed in completing the system. The third process involves investigating development tools, and the last step involves incorporating medical decision support systems. These stages can be summaries as follows (Chang and Lu 415-422).
Exploration and analysis of related literature
This stage involves collection of periodicals, internet articles, books, conference papers, theses for doctorates as well as masters, periodicals and all the materials pertaining to issues of medical decision support services as well as information from web services and their applicability. This is very essential for analysis, creation of theoretic and framework which forms the basis of this system (Chang and Lu 415-422).
Exploring ways to incorporate theories and framework
The next step involves investigation of ways that can be employed to integrate theories and frameworks explore above. These theories should have valid connections with existing tools required for completion of the system. This way, only correct framework and theories will be applied in creation of the system (Chang and Lu 415-422).
Exploring development tools
The following stage involves investigating right designs and development tools that can be applied in building the system. Such tools may include use of Microsoft Visual Studio 2003 or such other as applicable. This should be employed in conjunction with MS.NET with a view to developing the treatment decision support through web service technology. The systems also make use of java to develop decision support system for diagnosis. This is important in ensuring compatibility of the development tools to provide reliable results (Chang and Lu 415-422).
Incorporating medical decision support systems
The forth and final step involves incorporation of medical decision support system into web services. This is done to verify applicability of the system through application and integration of heterogenic medical support system. Results are important in determining reliability of the system as well as authenticity of explored literature and materials on medical decision support system along with their integration on web systems. The named steps must therefore be followed to achieve aim of this study (Chang and Lu 415-422).
Medical decision support systems
This system utilizes computers in analyzing pertinent information. While diagnosis is handled by interference and deductions; it also presents recommendations and treatment to those who provide care. This is usually done to enhance accuracy of diagnosis and to attain optimum treatment of medical decision problems. Evidences show that medical decision support systems have been available for a number of decades, during its initiation; the main focus was on creating systems that could employ data acquisition technology. It was then tried in data base management systems which were aimed at achieving better medical decision making. Internet application revolutionized medical decision support services as they could now incorporate it with the new technology to obtain a web-based medical decision support system. Among the systems created during this period for medical decision support include MYCIN and de Dombal, which were developed at the beginning stages. The others that were created after 1980s include Apache, DXplain, Problem-Knowledge Coupler, ILIAD, ATTENDING and QMR, among others (Chang and Lu 415-422).
These systems have provided valuable help to medical science and technology. Their significance can thus be divided into three categories, maintenance of medical security and public health, demand on medical field as well as impact on social cost. These points are very instrumental in building of medical decision support systems; they ensure it is in line with the goals of providing cheap, quality medical decision support to physicians using computer technologies such as web service technology. This objective is what drives development of the systems and have incorporated other medical personnel into providing quality medical decision services. From the above literature it is quite evident that building of medical decision support services has transformed throughout decades and the present usage employs contemporary technologies like internet and web services (Chang and Lu 415-422).
Web service
Data exchange was usually enabled through electronic data interchange which is also known as EDI. However, this method has proved expensive to use, and complex to command as it uses specific data formats and protocols that are very expensive to maintain. Furthermore, it was rigid and gave no allowance for expansion. This made it difficult to employ EDI in building medical decision support systems, and those devices made were also complex, costly and rigid. B2B data exchange was therefore difficult and straining for users, it also limited user as the skills required were complicated. The arrival of internet technology was timely as this made it easier to use other protocols such as TCP/IP and data formats like XML to transmit information thereby improving user interface as well as its integration in the web service technology. This created a new wave of B2B and B2C electronic data exchange businesses that are currently conventionally globalised (Chang and Lu 415-422).
Of special attention is B2B’s dominance in electronic business, in which there is incorporation of an e-marketplace model whereby third parties get the opportunity to provide transaction platforms for different buyers and sellers. Business dealings can now be done online through websites of B2B models, allowing various vendors to display their products globally. These transactions were done manually but this was solved by introduction of websites in which processes are automated. The rising challenge with automation is speed and efficiency. Competitiveness is based on speed of transactions as well as efficiency; websites that offer this are likely to gain higher traffics (Chang and Lu 415-422).
Web service refers to a programmable interface usually accessed in the internet. With availability of homogeneous communication protocols, web services can achieve automation and be accessed by every interested party. The use of web services involves (SOAP) simple object access protocols, (WSDL) web service description language, (UDDI) which is universal description, discovery and integration as well as XML which refers to extensible markup language. The system therefore involves incorporation of two services of different kinds in medical decision support systems for clients (Chang and Lu 415-422).
Components of web services
Web components serve three main primary roles when incorporated; these are web service requester, provider and the mediator between them. The table below shows web services components and their functions (Chang and Lu 415-422).
Service- Oriented architecture
This is a structure of IT that supports transformation of corporate business into repeatable missions of operations that allows access to the missions through the internet. Through incorporation of different technologies, it has the capability of giving an impression to the user that they are all located on the same screen. These can be grouped and integrated for different purposes and services as required. They are divided into three components, namely service provide, clients and SOA infrastructure. Web provides employ web services technology to change software functions to a service for use by the client. Another component is client, which gets services as required; the last component is SOA infrastructure which is more concrete with the simplicity of services. SOA infrastructure helps in expedition and repeated use (Chang and Lu 415-422).
Bayesian Theorem
Bayesian theory allows determination of the chance of an event. This is known as prior probability through probability principle after an in-depth understanding of the features of a population. When features of event are known and information is also availed (conditional probability), then prior probability can be modified to obtain posterior probability. The essence is integrating past along with current information in a limited environment to obtain a more accurate and reliable probability. This can be as shown below (Chang and Lu 415-422).
P (A/B) = P (B/A) * P (A)/ {[P (B/A) * P (A)] + [P (B/notA) * P (notA)]}
In the above formula, P (A) + P (notA) = 1, meaning that before a user determines posterior probability, new information should be obtained. After which, posterior probability can be obtained using Bayesian theorem (Chang and Lu 415-422).
Decision Trees
This method utilizes cost and effect of various stages to build decision trees and compute the effect and cost. They consist of decision branches as well as possible results which are usually formed via decision nodes. The latter refers to proposals while decision branches refer to the probabilities and their outcomes. Decision trees involves steps such as verifying of problems, setting up trees model, assessing probability of occurrence, verifying cost of every result, evaluating cost of decision nodes as well as conducting sensitivity analysis. The elements involved are of choice and opportunity. Decision tree is important in sensitivity analysis of information from Bayesian theorem (Chang and Lu 415-422).
System function
The study employs diagnosis decision support system written in Java language along with treatment decision support system created by MS.NET. These are then converted into web services by use of Apache AXIS. The two systems communicate to enhance exchange of information which is important in achieving its objectives. Use of MS.NET is important in enabling interaction and communication between web based interface and windows based applications. MS.NET refers to a software development platform that is independent and thus enables transparency of the network. On the other hand, AXIS employs use java as its primary development language and therefore has two operational versions, MS Windows and Linux. It (AXIS) refers to a web based platform that is founded on UNIX. It also has the ability to convert Java into WSDL and vice versa, with instantaneous provision of service. This makes it efficient and viable for use in this system (Chang and Lu 415-422).
The system has a user interface screen where they can input medical decision problems to obtain services required. After this has been obtained, (the application is based on Bayesian theorem) the user can then key aforementioned probabilities along with other associated conditions. When this information reaches the system, it will be transmitted to back-end web services where it will be computed. The results from this computation is then sent back in XML format to the front-end application program for Bayesian theorem after which it is presented to the users. Users can enter instructions for Bayesian application programs in the following ways: P (A) for prior probability, P (A/B) fro posterior probability, P (B/notA) for conditional probability and P (B/A) which stands for additional information (Chang and Lu 415-422).
On acquiring the posterior probability, users have the choice of transmitting those results to decision tree application, which conducts a further analysis on treatment decision. This is the stage at which users provide system information which must entail the treatment alternatives that they have chosen, result data as well as the associated symptoms. It is at this stage that users also decide on modifying the associated probability values. This information provided by users will be transmitted in a similar way as mentioned above to the back-end, where computation as well as analysis is done. The outcome is then transmitted to the front-end in which the structure of decision tree will represent expected values that are associated with decision choices and nodes of events as well as back-end computation respectively. The transmission will be presented on the front-end screen for user observation. Expected values that have been obtained will be compared by back-end web services and proper decision results sent to the front-end screen for users in XML format. The user will then conduct sensitivity analysis that would help in examining the whole dynamic model and try to lower medical risks (Chang and Lu 415-422).
Conclusion
This study aims to provide computer generated medical decision support system to users in conjunction with web services. It applies Bayesian theory in combination with web services as well as decision trees to put up a web-services-based which helps in decision support. Physicians key the available probabilities and obtain results (diagnosis) through computational analysis that employs Bayesian’s theorem. Through this theory, diagnosis is generated through a method which incorporates prior probability along with conditional probability to ascertain posterior probability. The system also utilizes decision trees which help users (clinical physicians) to incorporate decision logic as well as present information to output treatment decision recommendation and analysis (Chang and Lu 415-422).
This system can afford flexibility as opposed to the initial EDI. This is made possible by distinct packaging and cross-platform characteristics of web services via front-end interface. This helps in performing numerous jobs as well as enhancing efficiency and quality of medical decisions. Furthermore, these processes have the capacity to lower cost of the entire service (medical decision support system), improve its viability, speed as well as flexibility, thereby making it reliable and efficient (Chang and Lu 415-422).
Work Cited
Chang, Chung and Lu, Hsueh-Ming. “Integration of Heterogeneous Medical Decision Systems Based on Web services”. 2009 Ninth IEEE International Conference on Bioinformatics and Bioengineering. Taiwan.