Studies show that information needs that arise in clinical practices are frequently unmet. Some of the barriers to satisfying these needs include lack of up-to-date information resources, poor organization of available information, ignorance of the availability of relevant information, and lack of time for searching. Electronic resources such MEDLINE, full-text journal articles, decision support systems, and clinical Web sites begin to address users information needs but, however, these resources have the disadvantage of focusing on single functional areas such as decision support, literature searching, or Web exploration. Users whose information needs bridge these functional areas need to access each system separately and integrate the results for themselves. More desirable are applications that are designed to meet broad information needs and that integrate user information from many sources.
While searching for some literature on Decision Support systems, I came across an interesting paper written by William M. Detmer (Stanford University School of Medicine), G.Octo Barnett (Massachusetts General Hospital - responsible for development of DXplain) and William R. Hersh (Oregon Health Sciences University) about MedWeaver; where they explain the advantages and disadvantages of this system and its integration approach. The paper was written way back in 1997 when MedWeaver was only a prototype. Today, it is a full product and is being used as a Web-based facilitator that enables clinicians to make better-informed decisions. MedWeaver effectively integrates a powerful differential diagnosis tool with diverse information resources to generate personal answers.
MedWeaver
is a WWW application that integrates functions from a decision support
application (DXplain), a literature searching system (WebMedline), and
a clinical Web searching system (CliniWeb) using the UMLS [Unified Medical
language System] Metathesaurus for vocabulary translation. This system
demonstrates how application developers can design systems around anticipated
clinical information needs and then draw together the needed content and
functionality from diverse sources.
MedWeaver allows health professionals to:
- Enter clinical findings and retrieve a differential diagnosis
- Review a diagnostic profile for each disease in the differential diagnosis list
- Obtain an explanation as to why a disease was included on the differential diagnosis list
- Perform a powerful, assisted search of the world's medical literature in MEDLINE®, and
- Explore vetted, clinically relevant Websites related to each disease or condition
While the three stand-alone systems described above may be useful to clinicians, they are limited because they perform only a specific task-diagnosis of a patient, search of the literature, or search of the Web. MedWeaver is an application built using a model of clinical query management and demonstrates how functions from the three systems can be combined. The result is a decision support system that performs assisted searches of the medical literature and directs users to useful Internet sites.
MedWeaver
is a CGI application that resides at Stanford University, but gathers
information from systems in Portland, Oregon (CliniWeb); Boston, Massachusetts
(DXplain); and Bethesda, Maryland (UMLS). It is made up of several modules
as shown in figure below. It is written in PERL and runs on Unix servers.
The users interact with MedWeaver using a standard web browser. The web
server accepts the user requests and hands then off to the MedWeaver CGI.
The Interface Manager module decodes user requests and passes it to the
Query formulator module, which formulates appropriate queries using UMLS
translation services, and passes the query to the Retrieval manager. The
Retrieval manager module manages the retrieval of information from multiple
information sources based on this query. At the network level the retrieval
manager establishes and maintains connections with remote hosts using
agreed upon protocols and on an application level performs user authentication,
query submission and data retrieval. MedWeaver™ is a sophisticated decision
support tool that combines differential diagnosis, assisted literature
searching, and Web exploration. Using MedWeaver™, clinicians find answers
through two interactive modes: Differential Diagnosis and Disease Lookup.
The Differential Diagnosis mode allows entry of a patient case-complete
with patient demographics, signs, symptoms, and laboratory test results.
Figure below shows a sample screen of the look and feel and functionality.
When using MedWeaver, a Differential Diagnosis (DDx) (as above) is generated,
listing diseases by likelihood and prevalence. Each disease can be explored
through a Profile and an Explanation of why the disease or condition appears
in the differential diagnosis. MedWeaver's Differential Diagnosis (DDx)
screen shows diseases and conditions that match the symptoms and patient
characteristics entered. Profile, Medline, Web and Explain links provide
additional information.
- Enables clinicians to make better-informed decisions.
- Helps increase diagnostic quality and accuracy.
- The application provides accurate, relevant, accessible knowledge for clinicians.
- Allows
needs based as opposed to source-based design.
- Earlier applications were developed by taking an existing information source (e.g., MEDLINE) or by creating a new source (e.g. DXplain) and then developing functionality that maximized the potential of that individual source. But, with the advent of network-accessible information servers and integration technology, systems can be designed around the information needs of users and then use integration technology to retrieve the needed content and deliver the content at the appropriate point during the interaction with the user.
- Content creation and maintenance are distributed across institutions.
- Clinicians
quickly interact with medical knowledge in a user-friendly manner and
are guided through a unique presentation of focused content.
- The system has to rely on the quality of other's content, quality of wide area networks and reliability of information sources API's.
- The algorithm for vocabulary translation is insufficient when the base vocabulary term and the target vocabulary term don't share the same concept. Standardization of medical vocabulary and improved algorithms for these are concerns.
Note:
DXplain is a diagnostic DSS, which is capable of generating a differential diagnosis for a list of clinical finding, explain why a particular disease is triggered by a set of findings, or provide textbook information on a disease or finding. It stores medical knowledge and contains information on more than 2000 disease, 4700 clinical findings, and 65000 interrelationships. WebMedline is a CGI application that facilitates searching of medical literature via a web browser. CliniWeb is a retrieval system developed to help health practitioners find useful medical information on the WWW. UMLS Metathesaurus is one of the four knowledge sources in the National Library of Medicine's UMLS project. It is a collection of medical vocabularies tied together by the concepts they share.
- The full text of this paper can be found at http://www.people.virginia.edu/~wmd4n/AMIA97F.pdf.
- Product Information: http://www.unboundmedicine.com/index.htm