NETWORK BASED DIFFUSION OF KNOWLEDGE IN MEDICINE

University Degree Mathematical and Computer Sciences

Valerie Owunna, K0835950

MSC Information System with Management

CIM 951- Research Methods

Kingston University London

Abstract

The paper is based on the network diffusion of knowledge in medicine. In the medical field there has been a pressing need to integrate functions from disparate and diverse distribution information system to enhance knowledge dissemination.

The paper describes all the systems of the MedWeaver Architecture with emphasis on the how the MedWeaver integrates medical diagnostic decision support using DXplain, its literature searching using WebMedline and retrieval of clinically useful web site using CliniWeb. WebMedline make uses of UMLS Metathesaurus as a translator.

The paper has also covered the advantage and future challenges of using MedWeaver.

The conclusion summarises some main ideas found while preparing this paper and also the importance of using MedWeaver technology in the medical field.

1. Introduction

Young,M.A.(1967) stated that “the ultimate objective of most health education activities is to change individual, group, community, or societal behaviour.”

The traditional health and practices of the lay public, in cooperating the changing nature of health needs and the continuing appearance of research findings relevant to the developing practices and procedures to meet these needs argue for an increase concern with attempts to influence the attitudes and behaviours of health professionals have been stressed by the educator.

The frequent information needs in clinical practice are unmet (Smith, R.1996).Health sector has been interested in ensuring the rapid application of medical knowledge. The times spend between the releases of a new innovation (knowledge) and when it is used by other health practitioners has deprived people of the opportunity and benefits of these researches (Becker, M.H.1970).

In the past, applications were created by developers by using an existing information source (e.g.MEDLINE) or created a DXplain as a new source and then created functionality that amplified individual source. While, In this new era, developers are able to design systems around the users information needs to retrieve and deliver content at the right time during interaction with the user (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997).

The rapid growth in the medical field has encouraged expansion of knowledge, yet medical professionals are not aware of these relevant resources that are causing an immense change. It has been observed that the available information released in articles, researches, textbooks and medical journal are slowly disseminated.

Apparently, the obstacles behind this meeting these needs include Inadequate access to clinicians to update their information resources, Poor organization of accessible information, Insufficient amount of time for inquiry, and Lack of interest of clinicians on available important information in their field (Smith, R.1996) and (Demer,W.M. and Shortliffe,E.H.1995).

In order to achieve this goal is a combination of content, information science methods, and technology. Progress has been made by technologist in providing updated information to meet the needs of practitioners by using MebWeaver with a Common Gateway Interface (CGI) that integrates diagnostic decision support (DXplain), literature searching (WebMedline) and retrieval of clinically used websites (CliniWeb) and UMLS Metathesaurus translator (Detmer and Shortliffe 1997).

2. Background

2.1 Information Retrieval

The medical information retrieval systems are widely used in the medical libraries and on the internet. Information retrieval database is grouped as full text and bibliographic (Hersh,W.R.and Hickam,D.1995).The bibliographic database consist of original literature with indexer terms allocated by human indexer trained in medicine, to explore authoritative and relevant medical resources, assigning appropriate controlled vocabulary terms to them from Medical Subject Headings(MeSH) . Full text database consist of books, journals, e.t.c. that rely on index from words written in the entire indexed (Detmer and Shortliffe 1997) and (Hersh, W.R.and Hickam, D.1995).

Hersh,W.R.and Hickam,D.(1995) describe that the use of Medical Subject Headings (MeSH) terms to retrieve information is beneficial to librarians that are well trained to use such terms. Clinicians that are inexpert in the use of MeSH terms and as such do not show improved searching performance.

Detmer and Shortliffe (1997) argued in their paper that the web contains a vast collection of medical knowledge: journal, textbooks, e.t.c. recognizing the challenge faced by Clinician to locate this relevant information quickly. Therefore, the resulting controlled indexing term called Medical Subject Headings, have been established to aid the user to pick a specific query term. Thus, eliminating the frustration experienced by the clinicians in remembering the query language necessary to retrieve information, instead resources are navigated by subsequent links with a simplified interface.

2.2 Applications of Technology in Medicine

Medical professionals from different parts of the world are able to expand and disseminate medical knowledge among themselves over the years using the internet technology. Moreover the current internet has been able to meet the information needs of the clinicians by combining this widely distributed system components: technology, content of information and methods used in information science to meet these needs. Internet technology has contributed immensely to the success of medical professionals in medicine by interconnecting the Medweaver gateway with other applications to bring about a simple retrieval of current articles,journal and e.t.c.of knowledge in medicine, which merges these services of WebMedline, CliniWeb, DeXplain, UMLS Metathesaurus and MedWeaver. Initially, the retrieval systems in medicine made use of word-based demonstration approach.

2.2.1 CliniWeb

CliniWeb was developed by Oregon health Science University as a retrieval system therefore, making it easy for health professionals to locate useful, reliable and related information needs (Hersh,W.R.,et al.1997) and (Detmer and Shortliffe 1997).

This is a preview of the whole essay

2.2.1 CliniWeb

CliniWeb has been able to achieve this Practitioners needs by inventing a database of human filtered, customized web resources for practitioners that are concept-based automated indexing using Medical Subject Headings(MeSH).It functions in a way that as soon as a query (search) is initialized and placed in the database ,the database aids in accessing the request using any of these methods :searching interface that aids in mapping queries written in traditional natural language to Medical Subject Headings(MeSH) or browsing the interface that uses MeSH hierarchy including clinical resources related to the word and then viewing associated internet link based on the resources (Detmer and Shortliffe 1997). SAPHIRE is used by CliniWeb to map user’s free-text entries to MeSH. Semantic Probabilistic Heuristic Information Retrieval Environment (SAPHIRE) is a concept mapping engine used to identify the best techniques to index, retrieve and evaluate information retrieve resources in the biomedical domain of a unified Medical Language System (UMLS) provided by National Library of Medicine (Lindberg,D.A.,et al.1993).

In Addition, there are some challenges that will arise in the future use of CliniWeb systems which include: more storage space for information, maintenance of existing links, ability to locate and integrate new links to improve the mapping from natural language to controlled vocabulary terms.

2.2.2 WebMedline

WebMedline is a common Gateway Application (CGI) mediator, developed at Stanford University to enhance literature search in medicine through the web browser (Horres,M.M.,Starr,S.S., and Renford.B.L.1991). Detmer and Shortliffe (1997) pointed out that the Journal science and the Journal of Biological chemistry, literature review publication ACP journal club and the National Institutes of Health are accessible through WebMedline.The hypertext markup language (html) is used to display bibliographic data from MEDVYL MEDLINE.Medline database is executed using this same Language (html) to link full text articles on the web browser. For Instance: the use of hypertext markup language form to send a query to Medline database. Webline translates this query input into a legal Medline search statement, which is retrieved from the Medline and mark up the results in html (Demer,W.M. and Shortliffe,E.H.1995) (Detmer and Shortliffe 1997). MedWeaver and other remote application utilize a URL-based API owned by WebMedline to retrieve MEDLINE data (Deter,W.M.,Barnett,G.O.,and Hersh,W.R.1997).

Ovid Technologies, Inc. has a license for WebMedline and has been incorporated into Ovid Web Gateway Product.

2.2.3 DXplain

Massachusetts General Hospital developed the DXplain diagnostic decision support system (Barnett,G.O.,Cimino.J.J.,and Hoffer,E.P.1987) and (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997).The aim of DXplain is to produce a comprehensive list of likely medical diagnoses from a collection of clinical findings relevant to the developers. It has been observed that there is a major concern on interpretation of these results generated over the internet via a Telnet interface requires expertise of the practitioners, therefore it is made to available to Registered physicians. It possess an Applications Program Interface(API) permits DXplain functionality can be use in MedWeaver application (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997) and users can obtain information from other knowledge database.

Currently, it was restructured to serve as a knowledge base server has made created an opportunity for other developed web based front end e.g. WebDXplain and other applications on the internet that require diagnostic services. Pseudo-probabilistic algorithm is used to generate likely diagnoses in DXplain (Detmer and Shortliffe 1997).

2.2.4 UMLS Metathesaurus

The National library of Medicine developed a collection of medical vocabularies with similar concepts is called the unified medical language system’s(UMLS) Metathesaurus (Lindberg,D.A.,et al.1993).This is available on the internet through command-line,API or web interface.There are essential buttoms that appear in the UMLS Metathesaurus interface that aids in searching.UMLS Metathesaurus conserve its name,meaning,features, hierarchical context,and inter-term relating to the source vocabularies(Detmer and Shortliffe 1997).

3. Methods

The following stand alone systems that have been mentioned above are useful to the medical professionals but it has been observed that there are limitations using these systems. They provide information such as medical literature reviews, diagnosis of a patient and so on. An application called MedWeaver has been innovated to function as a clinical query management, in order to show how these wide distributed systems can be integrated and made available in ways that will meet the information needs of a practitioner. This, result to a system that supports decision making (Demer,W.M. and Shortliffe,E.H.1995) (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997).

3.1 MedWeaver Architecture

Stanford University developed a common gateway interface application that would enable knowledge transfer among three wide distributed systems called MedWeaver. The language format for MedWeaver is the Practical Extraction and Reporting Language (PERL) and it operates in a Unix Server. Over the decades, technology has been able to bridge this gap of slow diffusion of knowledge in the medical field by integrating these systems to communicate efficiently.

The Figure 1 consists of the components of a MedWeaver that initializes a session as soon as a request is receiver from a user (interface manager) using DXplain functions to locate this question term to a controlled vocabulary terms with a categorized list of likely search and summary attached to it. While, a WedMedline function will be used to retrieve information on the literature review and also locate clinically internet sites using CliniWeb functions (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997) and (Detmer,M.W. and Shortliffe,E.H. 1997).

Figure 1: MedWeaver Architecture. Users interact with MedWeaver using a Standard Web browser.When the Wed browser initializes the user’s request and hands them over to the MedWeaver CGI.Then, the MedWeaver’s modules decode user’s request(Interface Manager),formulate appropriate queries using UMLS Metathesaurus translation service (Query Formulator),retrieve information from remote systems(Retrieval Manager)which integrates and display the results(Interface Manager), (taken from:Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997).

3.1.1 Query Formulator

The duty of the query formulator is to translate the user’s request into a legal query of the remote information system. The interface on which the user makes a request is constructed using HTML format. For instance: If a user’s request is based on a single search centred around diagnosis, then a DXplain function can be implemented to achieve this request, but in a case where the request is not direct however, there is a need to merge at least two functionality of a separate systems(e.g., retrieve literature review WebMedline on the treatment used on a particular disease DXplain),the query formulator uses UMLS Metathesaurus to translate the vocabularies of the separate systems by sending a request to Methathesaurus for the concept code of the DXplain that matches the term in the base vocabulary .MedWeaver requests broader concepts for the root which finally request for target Metathesaurus term (Demer,W.M. and Shortliffe,E.H.1995) and (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997).

3.1.2 Retrieval Manager

The responsibility of a Retrieval Manager is to manage the task that relates to the retrieval of information from the information sources. In order for the retrieval manager to carry out this task, he maintains and establishes connection with the hosts using an acceptable protocol such as HTTP, Telnet or Sockets. The retrieval manager also authenticates the query from the user, submits the query and retrieves data. However, this can be achieved by the retrieval manager having a knowledge of the features of the remote system (e.g., error handling, style of interaction and format for the out), (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997).

3.1.3 Interface Manager

The interaction process that goes on between the user and interface manager is handled by the interface manager, which is used to integrate results of the processed incoming request and then displays it to the user. The interface manager sends the converted users request from HTML form into attributed value pairs to the query formulator. While the data extracted from the retrieval manager’s output is used to establish which output functions to provide, and then it marks up the selection output in HTML. The implicit model for clinicians’ information is rooted in the interface manager. This model helps to determine the functions available in the different stages of interaction with a user (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997) and (Detmer,M.W. and Shortliffe,E.H. 1997).

4. Results

The MedWeaver session initializes when a user writing clinical findings with the aim of receiving a differential diagnosis (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997) (Detmer,M.W. and Shortliffe,E.H. 1997). MedWeaver makes use of DXplain functions to map the clinical findings to DXplain controlled vocabulary terms and a list of ranked likely diagnoses are produces (Figure 2).MedWeaver,displays a list of the following for any diagnosis :

A summary of that disease (Figure 3 shows its DXplain function)
An Explanation as to why such diagnosis is seen in the list (Figure 4 shows its DXplain function)

An assisted search of the medical literature (Figure 5 shows its WebMebline function)

A site of clinically useful internet sites (Figure 6 shows its CliniWeb function)

Figure 2: Shows diagnoses generated by DXplain. MedWeaver uses a Dxplain function to retrieve a list of likely diagnosis for the clinical findings “adolescent female,” ”subacute(few days),” “lymph node enlargement,” and “sore throat.”From this page, users can retrieve a disease profile, view an explanation, perform an assisted search of medical literature, or retrieve a list of related Web resources (taken from:Detmer,M.W. and Shortliffe,E.H. 1997).

Figure 3: Portions of a disease profile for “infectious mononucleosis” provided by DXplain (taken from: Detmer, M.W. and Shortliffe,E.H. 1997) .

In order to carry out an assisted search using Medline, the query formulator retrieves the MeSH term from the Metathesaurus that is nearest to the DXplain disease name.so,it recovers from the aimed MeSH term its own concept definition, record of MeSH term , and the related MeSH term subheadings. While the Interface Manager makes use of these information to generate HTML form containing: the concept definition, subclasses of information about the disease, and the terms co-occurring (Figure 5).The required information is chosen by the user and sent in HTML form. The form is processed by the MedWeaver that invents a legal WebMedline query, citation from Medline with links to full text, which is then displayed in the hypertext.

In the stage of Query Formulation, MedWeaver creates a Boolean statement search made up of: the vital MeSH terms, an initial selected terms co-occurring and modifiers e.g.is a language and journal subset Abridged index Medicus(AIM).A mapped classes of information relating to MeSH subheadings (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997).

Figure 4: An explanation provided by DXplain for why “infectious mononucleosis appears on the list of possible diagnoses (taken from: Detmer, M.W. and Shortliffe,E.H. 1997).

Figure 5: An assisted search of the medical literature. MedWeaver first uses the UMLS Metathesaurus to find the closest MeSH term for the DXplain disease name “infectious mononucleosis.”Using this MeSH, it then retrieves from the Metathesaurus both the term definition and the terms definition and the terms that co-occur in the MEDLINE database. MedWeaver displays this information giving users the ability to limit searches to discrete classes of information such as “diagnostics tests” or prevention and control.”Once a user has specified the desired classes of information and co-occurring terms, MedWeaver generates a sophisticated query of the MEDLINE database and retrieves bibliographic citations and full- text links from WebMedline (taken from: Detmer,M.W. and Shortliffe,E.H. 1997).

Figure 6: List of clinically useful Internet sites provided by CliniWeb. MedWeaver finds the closest MeSH term for the DXplain disease “infectious mononucleosis,” requests from CliniWeb URLs indexed with the specific MeSH term, and then displays the URLs for the user (taken from: Detmer, M.W. and Shortliffe,E.H. 1997).

For example in figure 3 the diagnostic test choice is narrowed down to blood and radiography as the subheading. The user chose “Diagnostic tests” and the co-occurring term “Antibodies, Viral,” the Boolean search would result to “"(Assisted Literature Search: Infectious Mononucleosis/blood OR Assisted Literature Search: Infectious Mononucleosis /Radiography) AND Antibodies, Viral / AND Language English AND subset AIM." Also to retrieve clinically useful web URLs from CliniWeb, MeSH term can be detected using MedWeaver nearest to the generated DXplain name, as a result making use of Metathesaurus in the same way WebMedlline is carried out. While the request made by the Retrieval Manager is retrieved through CliniWeb API database records that will correspond to the MeSH term given. Furthermore, the interface manager will obtain from the CliniWeb record site name given, explanation on the site, and it’s URL which is then used to generate result on HTML page (Figure 6).

4. Discussion

Network based diffusion of knowledge in Medicine using World Wide Web architecture offers many advantages and disadvantages. The Common Gateway Interface(CGI) mediators, browser(www),and HTML has made it easy to access a disparate, widely distributed information systems that can be integrated by using the internet to produce a new application targeted to meet clinicians' information needs with a combination of content ,methods used in information science, and internet technology(Demer,W.M. and Shortliffe,E.H.1995).

There are potential advantages have been mentioned on how this architecture has helped in the diffusion of knowledge problem. Firstly, the advantage of developing a single, cross-platform application that retrieves information that is needs-based as opposed to source-based design. Secondly, the ability to shield clinician’s from the complexity of retrieval systems. There will be no need for users to know the query language of each information source. In alternative, developers can design user’s interfaces that closely correspond to the needs and ability of the user and therefore using the CGI as a mediator to map data entered into the interface of the language of each information resource (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997) and (Detmer,M.W. and Shortliffe,E.H. 1997). Thirdly, the integration of disparate and widely distributed information is achievable. Medical professional, can reach these information directly with information products that can merge useful components of previously separate information sources (i.e. being able to retrieve documents and at the same time reviewing patients data from the medical electronic record. fourthly, the content creation and maintenance are widely distributed across institutions. Also, individual organization may have limited resources and expertise for producing and maintaining medical content and also to produce libraries of information needed to meet individual information needs. Hence, merging the content distribution using integrated technology is a prudent approach to providing the content yet meeting users’ diverse information needs.

Lastly, the currency of information content, and quality of wide area networks has been improved .Thereby, eliminating the use of out dated textbooks, researches but having a direct access to the latest facts, opinion and summary from the source (Deter,W.M.,Barnett,G.O., and Hersh,W.R.1997) and (Detmer,M.W. and Shortliffe,E.H. 1997).

4.1 Future challenges

There still some limitations of this approach, they include:

4.1.1 Content

One of the future challenges is how the publishers can transfer printed materials to electronic-based for easy accessibility. Lately, publishers have been puzzled on how to break through this challenge knowing that they do not possess the expertise and internal processes to guide them in this development. The question is either they continue as content creators or follow the new demand in technological innovation in the sector of navigation, accessibility, distribution and search? In addition, there will be need to retool the editorial and document organizational methods to achieve this need.

4.1.2 Resource selection

The second challenge is on how to assist clinicians on the resource selection scheme for example: on what is the best available resources used to tackle a specific question, in order to eliminate the time spend on carrying out a continuous search due to the limited awareness of available resources. Presently, National Library of Medicine are working with the researchers to develop an efficient knowledge resource called Information Sources Map(ISM),which is embedded in the UMLS with the aim of creating source descriptions that can be shared and reused among Medical professionals. The Information sources Map contains a contents explanatory information, information procedures to connect and retrieve information from a source (Lindberg,D.A.,et al.1993) and ( Detmer,M.W. and Shortliffe,E.H. 1997).

4.1.3 Organization

Thirdly, the issue of organizing the output produced from these systems to meet users information needs in order to access material even when they are busy, is a challenge to be addressed .The various scheme of interest include: relevant ranking, relevant ranking based on users features, clustering, labelling of similar documents based on statistical correlations between: words, phases, and retrieved set concepts (Detmer,M.W. and Shortliffe,E.H. 1997).

4.1.4 Unified Interface

Fourthly, there is a need to disparate resources in a unified interface, thus giving the clinicians the opportunity to retrieve information from other resources possessing a different structure. However, the reason why there is a need for a unified interface is because the structure and retrieval paradigms of information sources vary widely.

4.1.5 Technology

In order to transfer knowledge in medicine, there is need to provide medical professional with connection to high speed data networks, thus giving them access to medical images, video, and others. However, there will be a likely chance that when electronic medical record and network based information are provided, it will boost the need for practitioners to invest more on network-based infrastructure (Detmer,M.W. and Shortliffe,E.H. 1997).

4.1.6 Patients Confidentiality

The issue of confidentiality and security has been a concern in the demand for high speed networks for the public health. In order to provide this needs, information that is been accessed within institution and across the globe by practitioners, maybe exposed to security threats and violations. Information on the user’s location, emails, and other information can be captured over the internet. A detail of a Patient’s medical condition, as entered into a retrieval system could cause a break into the confidentiality of the patient (Detmer,M.W. and Shortliffe,E.H. 1997).

5. Conclusion

The development of MedWeaver to meet clinicians’ information needs with systems that integrate information from disparate and diverse systems has been achieved with a combination of digital content, information science methods and technology. There are some challenges that need to be addressed in order to improve the existing knowledge base system.

Certainly, when these challenges are addressed, the practitioners would invest more on network infrastructure and this will ultimately boost the quality of medical care delivered to patients.

References

Barnett,G.O.,Cimino.J.J.,and Hoffer,E.P.(1987)DXplain.An evolving diagnostic decision-support system.Journal of the American Medical Informatics Association,258(1),67-74

Becker,M.H.(1970)Factors affecting diffusion of innovations among health professional.American Journal of Public Health(AJPH),60(2),294-304

Cimino,J.J.,Socratous,S.A.and Clayton,P.D.(1995)Internet as clinical information system:application development using the world wide web,Journal of the American Medical Informatics Association,2(5),273-284

Demer,W.M. and Shortliffe,E.H.(1995) A model of clinical query management supports integration of biomedical information over the world wide web.In Proceedings of the annual symposium on computer application in medical care,1995.Section on medical informatics,Stanford University school of medicine,pp.898-902

Detmer,M.W. and Shortliffe,E.H. (1997) Using the internet to improve Knowledge Diffusion in Medicine.Communications of the ACM,40(8), 101-108

Deter,W.M.,Barnett,G.O., and Hersh,W.R.(1997)MedWeaver:Integrated decision support,literature searching and web exploration using the UMLS Metathesaurus.In proceedings of the AMIA annual fall symposium,1997

Hersh,W.R.,et al.(1997) CliniWeb: Managing clinical information on the World-Wide Web. Journal of the American Medical Informatics Association, 4(1) 71 [Online]. Available at < > [Accessed 29 April 2009]

Hersh,W.R.and Hickam,D.(1995) Information retrieval in medicine: The sapphire experience. Journal of the American Medical Informatics Association, 46(10), 743-747

Horres,M.M.,Starr,S.S., and Renford.B.L.(1991)MELVYL MEDLINE:a library services perspective.Bull Med Libr Assoc 79(3),309-320

Lindberg,D.A.,et al.(1993) The unified medical language systems.Methods inf Med,32(4),281-291

Lindberg,D.A.,et al(1993) Use of medline by physicians for clinical problem solving.Journal of the American Medical Informatics Association, 269(24),3124-3129

Richardo,F.J.(2001)Hypertext and knowledge management.In proceedings of the 12th ACM conference on hypertext and hypermedia, 2001 ,pp.217-226

Smith,R.(1996)What clinical information do doctors need?Medical Publication of the Year(BJM)313(7064),1062-1068

Young,M.A.(1967)Review of research and studies related to health education.Health Education Monographs,25(7),1-6