Ebook: MEDINFO 2004

On behalf of the Scientific Program Committee we are delighted to welcome you to the triennial World Congress on Medical Informatics (Medinfo 2004), sponsored by the International Medical Informatics Association (IMIA), held in San Francisco, California, from September 7 to 11, 2004. It is hosted by the American Medical Informatics Association (AMIA) and has received a gratifying number of high -quality submissions, providing us all with an opportunity to share views on the future of our burgeoning field.

The main theme of this World Congress is building high performance organizations for enhancing health care, research, and education. The Scientific Program of Medinfo 2004 comprises 300 papers, and 450 posters presented in 7 tracks: Bioinformatics, Clinical Informatics, Education and Training, Enabling Technologies in Health Care, Human and Organizational Issues, Knowledge Management, and Public Health Informatics. In addition to the Keynote Presentation at the Opening Session, distinguished international experts are presenting semi-plenary sessions devoted to major themes of the Congress.

There will also be a full program of Workshops, Tutorials, and Working Group Meetings from IMIA and AMIA held throughout the Congress. The Scientific Program Committee met in December 2003 to finalize the selection of submissions, and authors were notified this past January. Based primarily on quality criteria, but also on the constraint on the number of sessions, the acceptance rate for 711 regular submitted papers was 42.2%. The acceptances levels ranged from a high of 49% for papers in the Knowledge Management track (out of 145 submitted) to a low of 34.1% (out of 28 submitted) for those in the Enabling Technologies track, which included imaging, signal processing and other innovative technologies. The largest number of papers were in the Clinical Informatics track (which also included papers submitted to the Patient Record and Nursing Informatics areas), with 245 submissions, of which 103 were accepted (40.8%), followed by Knowledge Management. The smallest number of submissions was in Bioinformatics, with 42 papers submitted, and 20 accepted (47.6%). Human and Organizational Issues had 76 submissions, of which 34 were accepted (44.7%), while Education and Training had 20 accepted out of 52 (38.5%), and Public Health Informatics had 24 out of 62 (38.7%). The distribution by continents of all 772 papers (including student papers and combined student/regular papers) was: Africa: 0.6%, Asia: 11.5%, Europe 35.5%, North America 47.5%, Oceania 2.7%, South America 2.1%. Due to the large number of excellent submissions, it was necessary to recommend that many submitted papers (30.2%) be presented as posters. While this may have been disappointing to some who had hoped to present, there is frequently an advantage to interested participants, because they can interact more directly, get more detailed feedback, and engage in more intense discussions. A major benefit of attending the World Congresses of Medical Informatics is the opportunity to meet old friends and make new ones, sharing ideas, and coming up with new ways of collaborating across the world on the medical informatics issues that excite us. We urge everyone to forge new links in the ever-widening network of IMIA organizations and professionals.

Medinfo 2004 features “Best Paper” and “Best Poster” awards based on selection by an international jury. A Student Paper Competition will also be part of the Congress, and awards will be presented to the top papers written by students in medical informatics programs. The Awards will be presented during the Closing Session of the Congress.

Medinfo 2004 comes at a propitious time in the historical development of our field: the last few years have seen a confluence of intellectual and scientific streams that drive our work. On the one hand, we are challenged by the overwhelming amount of genomic, proteomic, metabolomic, and other —omic data and knowledge that makes it not only attractive, but imperative for our researchers to tackle emerging informatics problems in genomic medicine. On the othex hand, life in our internet- or web-centric society makes us acutely aware almost daily of the ever increasing difficulties that medicine and public health have in managing the health problems of aging populations in developed nations, and of very young and often disease-ravaged populations in developing countries. These problems become ever more complex and unpredictable as technology, expanding populations, and globalization fuel an evolutionary explosion of microorganisms, including dangerous drug-resistant strains. Medical informatics offers incomparable experience in dealing with the clinical information systems that will be needed for genomic medicine, as well as the foundational tools of knowledge representation (such as ontologies) and information processing upon which these are based. Medical Informatics, is uniquely positioned to work on the fundamental methods for learning about the biomedical and health-related knowledge that will help us engineer our future within the constraints of our natural environment. An important part of this will require advancing the scientific, cognitive and social components of our informatics research and increasing our awareness of its impact on our professional practices. Information technology is deeply transforming the shape of organizations, the systems they use, and the knowledge they produce. Effective and efficient coordination of enterprise activities is important to the success of any organization. It is especially pressing for those in health care because of the strong and ever growing demand, and the need to keep up-to-date with costly and rapidly changing technologies. Information technology is increasingly expected to help deliver economies of scale and improved productivity, balancing costs with quality of health care. And, it is through the efforts of IMIA researchers, practitioners, and educators that some of the most advanced ideas and systems for both the science and the practice of biomedical informatics are being developed, disseminated, tested, and improved to face the ever-changing challenges of biomedical research and health care. We thank all contributors for joining in sharing your work through Medinfo 2004.

Medlnfo'04 Scientific Program Cochairmen

Mario Stefanelli (Italy)

Casimir A. Kulikowski (USA)

Clinical decision-making can be vastly improved with the availability of the right medical knowledge at the right time. This concept paper presents a knowledge management research program to (a) identify, capture and organize the tacit knowledge inherent within on-line problem-solving discussions between pediatric pain practitioners; (b) establish linkages between topic-specific pediatric pain discussions and corresponding published medical literature on children's pain available at PubMed–i.e. linking tacit expert knowledge to explicit medical literature; and (c) make these knowledge resources available to pediatric pain practitioners via the WWW for timely access to various modalities of clinical knowledge.

For many new medical research questions in heart surgery comprehensive and large data bases are essential. We discuss typical challenges for the integration of real-time and legacy data stored in multiple unconnected hospital information systems (HIS). Furthermore the HIS are often operated by autonomous departments whose data base structures are subject to occasional modifications.

We present a solution which integrates and consolidates all research relevant data in a data mart without imposing any considerable operational or maintenance contract liability risk for the existing HIS. The problems of partial consistency and partial redundancy in the data are discussed.

The data mart system serves multiple purposes: beside clinical reporting and quality assessment, the preparation steps for comprehensive studies are enormously simplified.

Since the widespread adoption of mammographic screening in the 1980's there has been a significant increase in the detection and biopsy of both benign and malignant microcalcifications. Though current practice standards recommend that the positive predictive value (PPV) of breast biopsy should be in the range of 25-40%, there exists significant variability in practice. Microcalcifications, if malignant, can represent either a non-invasive or an invasive form of breast cancer. The distinction is critical because distinct surgical therapies are indicated. Unfortunately, this information is not always available at the time of surgery due to limited sampling at image-guided biopsy. For these reasons we conducted an experiment to determine whether a previously created Bayesian network for mammography could predict the significance of microcalcifications. In this experiment we aim to test whether the system is able to perform two related tasks in this domain: 1) to predict the likelihood that microcalcifications are malignant and 2) to predict the likelihood that a malignancy is invasive to help guide the choice of appropriate surgical therapy.

Understanding how clinicians are using clinical information systems to assist with their everyday tasks is valuable to the system design and development process. Developers of such systems are interested in monitoring usage in order to make enhancements. System log files are rich resources for gaining knowledge about how the system is being used. We have analyzed the log files of our Web-based clinical information system (WebCIS) to obtain various usage statistics including which WebCIS functions are frequently being used. We have also identified usage patterns, which convey how the user is traversing the system. We present our method and these results as well as describe how the results can be used to customize menus, shortcut lists, and patient reports in WebCIS and similar systems.

In this paper, we present a method for aligning words based on a statistical model of word distribution similarity. The basis underlying our method is that there is a correlation between the patterns of word cooccurrences in texts of different languages. Using automatically downloaded pages from different medical web sites and a combined bilingual lexicon of general and medical terms as language sources, a similarity score is assigned to each proposed translated pair of words, based on the distributional contexts of these two words. We vary several parameters of the method. Experimental results confirm a positive effect of frequency, show that medical words are better handled than less specialized words, and do not evidence a clear influence of context window size. Future directions for improvement include working with very large, part-of-speech tagged corpora.

This paper describes the architecture of NewGuide, a guideline management system for handling the whole life cycle of a computerized clinical practice guideline. NewGuide components are organized in a distributed architecture: an editor to formalize guidelines, a repository to store them, an inference engine to implement guidelines instances in a multi-user environment, and a reporting system storing the guidelines logs in order to be able to completely trace any individual physician guideline-based decision process. There is a system “central level” that maintains official versions of the guidelines, and local Healthcare Organizations may download and implement them according to their needs. The architecture has been implemented using the Java 2 Enterprise Edition (J2EE) platform. Simple Object Access Protocol (SOAP) and a set of contracts are the key factors for the integration of NewGuide with healthcare legacy systems. They allow maintaining unchanged legacy user interfaces and connecting the system with whatever electronic patient record. The system functionality will be illustrated in three different contexts: homecare-based pressure ulcer prevention, acute ischemic stroke treatment and heart failure management by general practitioners.

Cancer researchers need to be able to organize and report their results in a way that others can find, build upon, and relate to the specific clinical conditions of individual patients. NCI Thesaurus This a description logic terminology based on current science that helps individuals and software applications connect and organize the results of cancer research, e.g., by disease and underlying biology. Currently containing some 34, 000 concepts – covering chemicals, drugs and other therapies, diseases, genes and gene products, anatomy, organisms, animal models, techniques, biologic processes, and administrative categories – NCI Thesaurus serves applications and the Web from a terminology server As a scalable, formal terminology, the deployed Thesaurus, and associated applications and interfaces, are a model for some of the standards required for the NHII (National Health Information Infrastructure) and the Semantic Web.

Situations managed by clinical practice guidelines (CPGs) usually correspond to general descriptions of theoretical patients that suffer from only one disease in addition to the specific pathology CPGs focus on. The lack of decision support for complex multiple-disease patients is usually transferred to computerbased systems. Starting from the GEM-encoded instance of CPGs, we developed a module that automatically generated IFTHEN- WITH decision rules. A two-stage unification process has been implemented. All the rules whose IF part was in partial matching with a patient clinical profile were triggered. A synthesis of triggered rules has then been performed to eliminate redundancies and incoherences. All remaining, eventually contradictory, recommendations were displayed to physicians leaving them the responsibility of handling the controversy and thus the opportunity to control the therapeutic decision.

Medical record linkage is becoming increasingly important as clinical data is distributed across independent sources. To improve linkage accuracy we studied different name comparison methods that establish agreement or disagreement between corresponding names. In addition to exact raw name matching and exact phonetic name matching, we tested three approximate string comparators. The approximate comparators included the modified Jaro-Winkler method, the longest common substring, and the Levenshtein edit distance. We also calculated the combined root-mean square of all three. We tested each name comparison method using a deterministic record linkage algorithm. Results were consistent across both hospitals. At a threshold comparator score of 0.8, the Jaro-Winkler comparator achieved the highest linkage sensitivities of 97.4% and 97.7%. The combined root-mean square method achieved sensitivities higher than the Levenshtein edit distance or longest common substring while sustaining high linkage specificity. Approximate string comparators increase deterministic linkage sensitivity by up to 10% compared to exact match comparisons and represent an accurate method of linking to vital statistics data.

Word-finding difficulty (anomia) is the most common linguistic deficit in dementia. It is often measured by picture naming tasks as naming a picture taps all the major processes in word production, i.e., activation of a concept, retrieval of lexical-semantic information on that concept, retrieval of the corresponding word form and articulation. Naming and naming errors have extensively been simulated by neural network models of lexicalization (see e.g. [1,2]). A common feature of these models is that they are static, i.e. non-learning. However, naming is a dynamic process that changes as a function of normal learning or re-learning after neural damage. These important patterns cannot be caught by the static models of lexicalization. Therefore we have developed a learning model of lexicalization based on multi-layer- perceptron (MLP) neural networks. We tested the model by fitting it to the naming data of 22 Finnish-speaking dementia patients and 19 neurologically intact control subjects. The tests showed an excellent fit between the model’s and the subjects naming response distributions. Thus our model seems be suitable to simulate naming disorders of dementia patients.

The U.S. National Library of Medicine (NLM) has created a metasearch engine called the NLM Gateway at the URL “gateway.nlm.nih.gov”. The Gateway allows the user to issue one search that takes place on multiple NLM retrieval engines. A composite result set is presented in several categories of information: journal citations; books, serials and audiovisuals; consumer health; meeting abstracts; and other collections.

Clinical protocols and guidelines are widely used in the medical domain to improve disease management techniques. Different software systems are in development to support the design and the execution of such guidelines. The bottleneck in the guideline software developing process is the transformation of the textbased clinical guidelines into a formal representation, which can be used by the execution software. This paper introduces a method and a tool that was designed to provide a solution for that bottleneck The so-called Guideline Markup Tool (GMT) facilitates the translation of guidelines into a formal representation written in XML. This tool enables the protocol designer to create links between the original guideline and its formal representation and ease the editing of guidelines applying design patterns in the form of macros. The usefulness of our approach is illustrated using GMT to edit Asbru protocols. We performed a usability study with eight participants to examine the usefulness of the GMT and of the Asbru macros, which showed that the proposed approach is very appropriate to author and maintain clinical guidelines.

This paper describes a method for using Semantic Web technologies for sharing knowledge in healthcare. It combines deductive databases and ontologies, so that it is possible to extract knowledge that has not been explicitly declared within the database. A representation of the UMLS (Unified Medical Language System) Semantic Network and Metathesaurus was created using the RDF standard, in order to represent the basic medical ontology. The inference over the knowledge base is done by the TRI-DEDALO System, a deductive database created to query and update RDF based knowledge sources as well as conventional relational databases. Finally, an ontology was created for the Brazilian National Health Card data interchange format, a standard to capture and transmit health encounter information throughout the country. This paper demonstrates how this approach can be used to integrate heterogeneous information and to answer complex queries in a real world environment.

The National Library of Medicine (NLM) produces annual editions of the Medical Subject Headings (MeSH®). Translations of MeSH are often done to make the vocabulary useful for non-English users. However, MeSH translators have encountered difficulties with entry vocabulary as they maintain and update their translation. Tracking MeSH changes and updating their translations in a reasonable time frame is cumbersome. NLM has developed and implemented a concept-centered vocabulary maintenance system for MeSH. This system has been extended to create an interlingual database of translations, the MeSH Translation Maintenance System (MTMS). This database allows continual updating of the translations, as well as facilitating tracking of the changes within MeSH from one year to another The MTMS interface uses a Web-based design with multiple colors and fonts to indicate concepts needing translation or review. Concepts for which there is no exact English equivalent can be added. The system software encourages compliance with the Unicode standard in order to ensure that character sets with native alphabets and full orthography are used consistently.

The application of principles and methods of cybernetics permits clinicians and managers to use feedback about care effectiveness and resource expenditure to improve quality and to control costs. Keys to the process are the specification of therapeutic goals and the creation of an organizational culture that supports the use of feedback to improve care. Daily feedback on the achievement of each patient's therapeutic goals provides tactical decision support, enabling clinicians to adjust care as needed. Monthly or quarterly feedback on aggregated goal achievement for all patients on a clinical pathway provides strategic decision support, enabling clinicians and managers to ident problems with supposed “best practices” and to test hypotheses about solutions. Work is underway at Vanderbilt University Medical Center to implement feedback loops in care and management processes and to evaluate the effects.

Computer simulation enables system developers to execute a model of an actual or theoretical system on a computer and analyze the execution output. We have been exploring the use of Petri Net (PN) tools to study the behavior of systems that are represented using three kinds of biomedical models: a biological workflow model used to represent biological processes, and two different computer-interpretable models of health care processes that are derived from clinical guidelines. We developed and implemented software that maps the three models into a single underlying process model (workflow), which is then converted into PNs in formats that are readable by several PN simulation and analysis tools. We show how these analysis tools enabled us to simulate and study the behavior of two biomedical systems: a Malaria parasite invading a host cell, and patients undergoing management of chronic cough.

Previous papers have argued for the existence of three different models in many clinical information systems — for the medical record, for inference in guidelines, and for concepts and reusable facts. This paper presents a principled approach to deciding which information belongs in each model based on the nature of the queries or inference to be performed: necessary or contingent, open or closed world, algorithmic vs heuristic. It then discusses an important class of systems – “ontologically indexed knowledge bases” – and issues of metadata within this framework

Query and interpretation of time-oriented medical data involves two subtasks: Temporal-reasoning–intelligent analysis of timeoriented data, and temporal-maintenance–effective storage, query, and retrieval of these data. Integration of these tasks into one system, known as temporal-mediator, has been proven to be beneficial to biomedical applications such as monitoring, therapy, quality assessment, visualization and exploration of timeoriented data. One potential problem in existing temporal-mediation approaches is lack of sufficient responsiveness when querying or continuously monitoring the database for complex abstract concepts that are derived from the raw data, especially regarding a large patient group. We propose a new approach: the knowledge-based time-oriented active database, a temporal extension of the active-database concept, and a merger of temporal reasoning and temporal maintenance within a persistent database framework. The approach preserves the efficiency of databases in handling data storage and retrieval, while enabling specification and performance of complex temporal reasoning using an incremental-computation approach. We implemented our approach within the Momentum system. Initial experiments are encouraging; an evaluation is underway.

Growing complexity of diagnostic tests, combined with increased workload, stringent laboratory accreditation demands, continuous shortening of turn-around-time and budget restrictions have forced laboratories to automate most of their iterative tasks. Introduction of artificial intelligence by means of expert systems has gained an important place in this automation process. Different parts of clinical laboratory activity can benefit from their implementation and the present project deals with one aspect, namely the clinical interpretation of diagnostic tests. This paper describes how j.MD, a new Java based expert system shell, was used to reprogram the expert system for interpretation of amylase isoenzyme patterns that has been in use for many years in our laboratory, and that was originally programmed in Pro.MD, a Prolog based expert system shell. One of the most important advantages of the j.MD system is its bidirectional link with the laboratory information system. This project shows how expert systems for the interpretation of complex diagnostic tests that demand specific expertise can become an integrated part of the automated clinical chemistry lab.

With the information on the World Wide Web and in specialized databases exploding, researchers and physicians are in dire need to browse efficiently though the large corpus of information resources in their field of interest. The focus is not any longer to find everything related to your interest, but it shifts to zooming in, based on context and expanding again in neighboring knowledge domains. This paper describes an attempt to develop a completely new, interactive way of browsing distributed corpora of information without the need for multiple different queries in different information resources.

Classical search engines generally treat search requests in isolation. The results for a given query are identical, and do not automatically take on board the context in which the user made the request. The system described here explores implicit contexts as obtained from the document that the user is reading. The new approach merges the searching and browsing into one combined “read-and-search” mode and alleviates the shift users are normally forced to between searching and reading.

Diffusing knowledge management practices within an organization encourages and facilitates reuse of the institution’s knowledge commodity. Following knowledge management practices, the Eskind Biomedical Library (EBL) has created a Digital Library that uses a holistic approach for integration of information and skills to best represent both explicit and tacit knowledge inherent in libraries. EBL's Digital Library exemplifies a clear attempt to organize institutional knowledge in the field of librarianship, in an effort to positively impact clinical, research, and educational processes in the medical center.

A dynamic decision analytic framework using local statistics and expert's opinions is put to study the cost-effectiveness of colorectal cancer screening strategies in Singapore. It is demonstrated that any of the screening strategies, if implemented, would increase the life expectancy of the population of 50 to 70 years old. The model also determined the normal life expectancy of this population to be 76.32 years. Overall, Guaiac Fecal Occult Blood Test (FOBT) is most cost effective at SGD162.11 per life year saved per person. Our approach allowed us to model problem parameters that change over time and study the utility measures like cost and life expectancy for specific age within the range of 50- 69 through to 70 years old.