ARTIFICIAL INTELLIGENCE IN MEDICINE (AIME-93) PROGRAMME ********************************************************************** Welcome to AIME 93 The AIME 93 conference is organized by the European Society for Artificial Intelligence in Medicine and is the fourth in the Societies biannual series of international conferences. These conferences have become a central forum both for researchers that are developing new methods for construction of medical decision support systems and for researchers that are using these methods to construct systems that are providing medical decision support for a wide range of medical problems. The scientific programme includes about 50 presentations, organised around the central themes: * Model Based Reasoning * Strategies for Medical Reasoning and * Knowledge Acquisition, Representation and Learning. To emphasize the importance of applications special attention has been given to the poster sessions and to the demonstrations. The Program Committee thanks all the contributors to the conference for making it possible to offer such a scientifically interesting programme. In addition a tutorial programme is offered as the first activity at the conference on October 3rd. The conference takes place in Munich, a city of tradition and culture - and the city of the famous "Oktoberfest", which ends on October 3'rd. The prospects of holding a very interesting conference, where old friendships are strengthened and new friendships are made, are very fine indeed. Once again, welcome to Munich. On behalf of the Organizing and Programme Committees Steen Andreassen AIME and AIME 93 The European Society for Artificial Inteligence in Europe was established in 1986 to foster fundamental and applied research in artificial intelligence and symbolic information processing techniques for medical care and medical research. AIME also assists industry in identifying high quality medical products which exploit these techniques. A major AIME activity was and will be a biannual series of international conferences: 1987 Marseille 1989 London 1991 Maastricht 1993 Munich 1995 Pavia These conferences show current scientific results presented in oral contributions, posters, and demonstrations. An important part are the tutorials which are methodological as well as user oriented. AIME 93 will provide invited papers contributed papers contributed posters system demonstrations round table discussions tutorials to introduce newcomers and discuss advanced topics in the following subject areas: knowledge representation neural networks uncertainty in medical knowledge natural language understanding knowledge based image interpretation other areas relevant to AI in medicine Expected Audience The conference is intended for clinicians, computer scientists, medical informaticians, statisticians, and other medical professionals interested and working in the subject areas. Preliminary Scientific Programme Sunday 3rd October Tutorials 09:30 - 13:00 Morning session Introductory tutorial Jeremy Wyatt, London Evaluating medical decision-aids: why, what, where and how ? Advanced tutorial Kai von Luck, Bjoern Owsnicki-Klewe, Hamburg Semantic Networks and Term Description Languages for Knowledge Representation 14:00 - 17:30 Afternoon session Introductory tutorial William J. Long, Cambridge Designing Mechanisms to Capture Reasoning in a Medical Domain Advanced tutorial Michael G. Kahn, St. Louis Time Traps in Medical Decision Support Systems Monday 4th October 08:30 - 09:00 Welcome address 09:00 - 10:30 Model Based Reasoning I Ironi, L., Cattaneo, A., Stefanelli, M. (Pavia) A tool for pathophysiological knowledge acquisition Kamber, M., Shinghal, R., Evans, A.C., Collins, D.L., Francis, G.S. (Montreal) Knowledge-Based Interpretation of Magnetic Resonance Images: Detecting Multiple Sclerosis Lesions Arana, I., Hunter, J. (Aberdeen) Applying Qualitative Reasoning to Commonsense Physiology 11:00 - 12:30 Model Based Reasoning II Hejlesen, O.K., Andreassen, S., Andersen, S.K. (Aalborg) Implementation of a learning procedure for multiple observations in a Diabetes Advisory System based on causal probabilistic networks Barahona, P., Cruz, J., Nascimento, S. (Lisboa) Reasoning with causal networks at variable depths Oppel, U.G., Hierle, A., Janke, L., Moser, W. (Munich) Transformation of Compartmental Models into Sequences of Causal Probabilistic Networks 14:00 - 14:45 Keynote Speaker: Michael Kahn In Pursuit of Time's Arrow: Reasoning about Time in Medicine 15:15 - 16:15 Posters with Demonstrations Herpers, R., Kattner, H., Sommer, G. (Munich) GAZE: An Attentional Mechanism for the Extraction of Prominent Facial Features Schroeder, J.P., Henne, P., Kuhlmann, W.D. (Bonn) Clinical decision-support in the field of TETANUS Serology using an associative storage model implemented in LISP Englmeier, K.-H., Haubner, M., Foersterer, H., Fink, B.K., Perzl, W., Waltram, R. (Munich) Computer-Assisted Medical Decision Support by Analysis and Visualization of Multimodal Images Post, W.M., Sramek, M., Koster, R.W., Zocca,V. (Amsterdam) FreeCall: A System for Ambulance Dispatching Stausberg, J., Diedrich, T., Wormek, A., Hucklenbroich, P. (Munich) A Model of Patient Related Data and a Structure of Medical Terms for Knowledge-based Systems in Medicine. Oppel, U.G., Hierle, A., Janke, L., Moser, W. (Munich) Metabolic Compartmental CPNs, an Improved Description of Metabolic Processes by CPNs from Compartmental Models. Hejlesen, O.K., Andreassen, S., Andersen, S.K. (Aalborg) DIAS: A Diabetes Advisory System based on causal probabilistic networks Lipsanen, M., (Abo), Andreassen, S. (Aalborg), Falck, B., Halonen, P. (Turku) A Prototype Decision Support System for Diagnosing Polyneuropathies Molino, G., Nicolosi, E., Molino, F. (Torino) CARE - A Knowledge-Based System for Assisting the Physician in the Patient Management Gierl, L., Schmidt, R., Pollwein, B. (Munich) ICONS: Cognitive basic functions in a case-based consultation system for intensive care Goos, K., (Wuerzburg), Schewe, S. (Munich) Case Based Reasoning in Clinical Evaluation 16:30 - 18:00 Strategies for Medical Reasoning - ICU Applications Lillehaug, S.-I. (Tromso) ILE-VT: An Intelligent Learning Environment in Ventilator Therapy Andreassen, S., Egeberg, J., Schroeter, M.P., Andersen, P.T. (Aalborg) Oxygen Status Model for the Ventilatory and Circulatory Contribution to Tissue Oxygenation Miksch, S., Horn, W., Popow, C., Paky, F. (Vienna) VIE-VENT: Knowledge-Based Monitoring and Therapy Planning of the Artificial Ventilation of Newborn Infants Tuesday 5th October 08:30 - 10:00 Strategies for Medical Reasoning - Patient Management Ovalle, A., Hugonnard, E., Garbay, C. (Grenoble) Medical system design and knowledge acquisition using cooperating intelligent agents - A case study for breast cancer diagnosis Post, W.M., Koster, R.W., Zocca, V., Sramek, M. (Amsterdam) Cooperative Medical Problem Solving Molino, G., Nicolosi, E., Molino, F. (Torino) CARE - A Knowledge-Based System for Assisting the Physician in the Patient Management 10:30 - 12:30 Knowledge Representation - Protocols Gordon, C., Jackson-Smale, A. (London), Herbert, I. (Birmingham), Renaud-Salis, J.-L. (Bordeaux) Care Protocols and Healthcare Informatics Beckers, W.P.A., de Vries Robb, P.F., Zanstra, P.E., Zwaard, A.M. (Nijmegen) Generic GP-guidelines and reasoning Quaglini, S., Bellazzi, R., Stefanelli, M., Locatelli, F., (Pavia) Sharing and reusing therapeutic knowledge for managing leukemic children Mller, L.M. (Eindhoven), Hasman, A., Blom, J.A. (Maastricht) Expert Systems and Mathematical Simulation 14:00 - 14:45 Keynote Speaker: Jan van Bemmel Criteria for the Acceptance of Decision Support Systems by Clinicians 15:15 - 16:15 Poster Presentations Englmeier, K.-H., Haubner, M., Fink, B.K., Fink, U. (Munich) Model Based Image Interpretation of Spiral CT Scans of the Abdomen Blonda, P., Carella, A., De Blasi, R., Dicuonzo, F., La Forgia, V., Milella, D., Pasquariello, G., Satalino, G., (Bari) A Neural Network Modular System for Object Classification in Brain MR Images von Wangenheim, A., Vince, G.H., Kolles, H., Richter, M.M., Feiden, W. (Kaiserslautern, Homburg) Grading of Gliomas in Stereotactic Biopsies with Neuronal Networks Herpers, R., Rodax, H., Sommer, G. (Munich) A Neural Network identifies Faces with Morphological Syndromes Schulz, B., Pelikan, E., Silny, J., Rau, G. (Aachen) Detection of characteristic high resolution ECG-changes using self-organizing maps Kindler, H., Densow, D., Fliedner, T.M., (Ulm) A Knowledge-Based Advisor Supporting to Deal with Rare Diseases Keller, F., Erdmann, K., Giehl, M., Czock, D. (Berlin) Pharmneph: Computer-system for drug dosage adjustment in renal failure Safran, E., Borst, F., Thurler, G., Pittet, D., Lovis, Ch., Rohner, P., Auckenthaler, R., Scherrer, J.-R. (Geneva) An evolutive alert system for diagnosis of nosocomial infection based on archiving of distributed databases Fiore, M., Sicurello, F., Vigano, R., Pizzi, R., Bandini, S. (Milano) A knowledge-based system to classify and diagnose HIV-pneumonias Le Blanc, B., Palmer, G. (Bordeaux) AntibES: An Expert System for Antibiotic prescription based on fuzzy logic Picart, D. (Brest) Qualitative modelisation of biochemical interactions using a network of automations - application to the metabolism of purines and related compounds 16:30 - 18:00 Knowledge Representation - Terminology Alpay, L., Baud, R., Rassinoux, A.-M., Wagner, J., Lovis, C., Scherrer, J.-R. (Geneva) Interfacing Conceptual Graphs (CG) and the Galen Master Notation (MT) for medical knowledge representation and modelling Bernauer, J., Goldberg, H. (Hildesheim) Compositional Classification Based on Conceptual Graphs Lovis, C., Baud, R., Alpay, L., Rassinoux, A.-M., Wagner, J., Scherrer, J.-R. (Geneva) Needs and constraints in modelling medical knowledge using semantic networks: A medical point of view Wednesday 6th October 08:30 - 11:00 Knowledge Acquisition, Representation & Learning Brendel, O., Widdig, R., Piron, F., Freiburg, Graf von Keyserlingk, D. (Aachen) Knowledge Representation and Modelling in HYBRIKON Schreiber, A. Th., van Heijst, G. (Amsterdam), Lanzola, G., Stefanelli, M. (Pavia) Knowledge Organisation in Medical KBS Construction Heller, B., Schlegelmilch, S. (Bielefeld) A Framework for Modular Knowledge Bases in the Domain of Hypertension Diseases Vargas, J.R., Bologna, G., Appel, R.D., Hochstrasser, D.F., Pellegrini, C. (Geneva) Classification of protein patterns using neural networks: pixel based versus feature based approach Waschulzik, T., Quandt, K., Lewis, M., Hrmann, A., Engelbrecht, R., Brauer, W. (Munich) Evaluation of an epidemiological data set as an example of the application of neural networks to the analysis of large medical data sets 11:30 - 12:30 Poster Presentations Escalada-Imaz, G., Jaureguizar-Nunez, J. (Gerona), Pastor-Duran, X., Fita-Rodriguez, G. (Barcelona) Interpreting Physiopathological States under Environment Temporal Restrictions Gross, H., Fhring, U. (Berlin) Digital lung mechanics simulator as a validation tool Morik, K. (Dortmund), Potamias, G., Moustakis, V., Charissis, G. (Heraklion) Model based learning support to knowledge acquisition: A clinical case study Leaning, M.S., Samuel, P., Austin, T., Hassan, T., Modell, M., Petterson, D., Rubens, R. (London) Clinical Validation of a General Methodology for decision-support systems Goldberg, S.I., Tsibulkin, E.K., Makhanek, A.O., Meshalkin, L.D. (Yekaterinburg, Sanct-Petersburg, Moscow) Employment of Dr. Watson Type System (Dr. WTS) in medicine on the example of supporting decision-making by a physician of the Regional Children Reanimation Consultative Centre Riou, Ch., le Beux, P., Lenoir, P., Leclet, D., (Rennes) Models for medical knowledge representation and medical reasoning in a CAI system Petkoff, B. (Bad Oeynhausen) A framework for computer assisted medical decision making Puppe, B. (Wuerzburg) Building a Medical Knowledge Base: Tricks Facilitating the Simulation of the Experts Reasoning Popper, M. (Bratislava) On higher level operational facets of diagnostic expert systems Mommaerts, J.-L. (Antwerp) Object-oriented mentality: the most suited paradigm for medical knowledge-based systems 14:00 - 16:00 Panel Discussion Perspectives and priorities for R & D programs in AIM Artificial intelligence and knowledge-based systems are important research areas in R & D programs in advanced informatics in medicine in Europe. Given the current context of increasing pressure towards more applied research, the panel will address the perspectives and priorities for R & D in these areas for the next 5 to 10 years. Panel participants: Mario Stefanelli, Jens Christensen, Niels Rossing, Jean-Louis Renaud-Salis, Jeremy Wyatt, Wilhelm van Eimeren. Tutorial Details Jeremy Wyatt, London Evaluating medical decision-aids: why, what, where and how ? Abstract Medical decision-aids are decision-support systems which provide patient-specific advice [1]. Doubts about their safety and efficacy are major impediments to clinical acceptance, and will only be allayed by careful evaluation of their structure, function and impact [2]. Funding bodies and journal referees are also exerting pressure on developers to evaluate systems, even prototypes built solely to explore research issues. This tutorial will review the need for and problems of evaluation, discuss what to measure, and explain how to make these measurements objectively in both laboratory and field settings. The instructor will use published reports [eg. 3, 4] and his practical experience of conducting and interpreting evaluations [eg. 5, 6] to illustrate these issues. Intended audience The intended audience is those developing or using medical decision-aids or who wish to interpret evaluation reports or conduct evaluations. No experience of evaluation methods will be assumed. Attendees will participate in a short exercise and will receive detailed reading material. References 1. Wyatt, J.: "Computer-based knowledge systems" The Lancet 1991; 338:1431-1436. 2. Wyatt, J., Spiegelhalter, D.: "Evaluating medical expert systems" Medical Informatics 1990; 15: 205-217. 3. Willems, J.L. et al:. "The diagnostic performance of computer programs for the interpretation of ECGs" New Engl J Med 1991; 325:1767-1773. 4. Wellwood, J. Spiegelhalter, D.J. and Johannessen, S.: "How does computer-aided diagnosis improve the management of acute abdominal pain ?" Annals Royal Coll Surg of England 1992; 74: 140-146. 5. Wyatt, J.: "Lessons learned from the field trial of ACORN, an expert system to advise on chest pain" In: Barber, B., Cao, D., Qin, D. (eds.) Proc. 6th MedInfo, Singapore. North Holland 1989: 111-115. 6. Wyatt, J., Spiegelhalter, D.: "Field trials of medical decision-aids: potential problems and solutions" In Clayton P (ed). Proc. 15th SCAMC, Washington 1991. New York: McGraw Hill Inc. 1991: 3-7. Tutor Jeremy Wyatt MRCP DM (Oxon) manages clinical software development at the Imperial Cancer Research Fund, London and oncology patients in Oxford. He recently spent a year at Stanford University where he taught evaluation methods and participated in the design of a number of experiments. He has given tutorials at various meetings including the AMIA Spring meeting in 1992 and MIE 1993. He is currently writing a book with Prof. Charles Friedman (Univ. of North Carolina) entitled "Evaluation & Research Methods in Medical Informatics" for the Springer Verlag "Computers in Medicine" series. Kai von Luck, Bjoern Owsnicki-Klewe, Hamburg Semantic Networks and Term Description Languages for Knowledge Representation Abstract The aim of this tutorial is to give an overview of recent research on knowledge representation by logic-based semantic network formalisms. After a short introduction to the field of knowledge representation in AI and a description of semantic networks as representation schemes for psychological models of human memory, Term Description Languages (TDLs) will be introduced as an updated form of semantic networks. TDLs represent a set of formalisms, best known to AI through the modelling language KL-ONE, which can be regarded as one of the best understood formalisms for knowledge representation. After introducing TDLs and their logical basis and the resulting derivation, the expressiveness of specific TDLs versus their inferential costs will be discussed. Examples of modelling by means of a TDL show how to deal with such formalisms and illustrate their facilities and limits. The tutorial will be completed by a presentation of implemented TDLs. Intended Audience: The intended audience is those interested in logic-based approaches for knowledge representation. Basic knowledge of AI methods in general and of knowledge representation in particular is welcomed but is not, however, necessary. Tutors Kai von Luck was formerly with IBM and a member of the LILOG project, which is Germany's largest project in the field of knowledge representation. Bjoern Owsnicki-Klewe formerly worked in the Philips research center. Both are professors at the department of electronic engineering and computer science at the Fachhochschule Hamburg. William J. Long Designing Mechanisms to Capture Reasoning in a Medical Domain Abstract Whether one uses Bayesian nets, criteria tables, qualitative or quantitative simulation, neural nets, or something else, the mechanism never quite matches the knowledge available in the target medical domain. Still, there are important features of each domain that frame the essential distinctions. These may be captured well by a particular mechanism or combination of mechanisms. The challenge for the designer is to analyze the domain knowledge to determine what kinds of relationships support the essential reasoning and therefore what reasoning mechanisms can utilize those relationships and implement the reasoning. In the tutorial, we will look at the most popular reasoning mechanisms to determine what kinds of domain relationships they utilize, how they have been used, and what domain characteristics should be considered in selecting them. Since the armamentarium is hardly complete, we will also look at some of the relationships that exist in medical domains and discuss what aspects are difficult to represent and utilize. References 1. William J. Long: "Medical Diagnosis Using a Probabilistic Causal Network,'' Applied Artificial Intelligence, 3:367-383, 1989. 2. William J. Long, Shapur Naimi and M. G. Criscitiello: "Development of a Knowledge Base for Diagnostic Reasoning in Cardiology,'' Computers in Biomedical Research, 25:292-311, 1992. 3. William J. Long, John L. Griffith and Harry P. Selker, and Ralph B. D'Agostino, "A Comparison of Logistic Regression to Decision Tree Induction in a Medical Domain,'' Computers in Biomedical Research, 26:74-97, 1993. Intended audience The primary audience for this tutorial is someone considering implementing a medical expert system. The discussion will be at a fairly general level, concentrating on characteristics rather than details, so it will also be useful to people who want a better understanding of the strengths and weaknesses of different kinds of systems. Tutor William Long has been involved in the design and implementation of medical expert systems at the MIT Lab for Computer Science since 1977. Over the past ten years he has been the designer and implementor of the Heart Failure Program, working with a team of cardiologists. The program combines aspects of Bayesian reasoning, constraint based reasoning, quantitative simulation, and case based reasoning. The tutorial will draw on this experience as well as experience with other reasoning methodologies. Michael G. Kahn, St. Louis Time Traps in Medical Decision Support Systems Abstract To derive meaningful conclusions in a changing medical setting, medical decision-support systems must represent and reason about the temporal nature of the clinical environments they attempt to model. Designers of medical decision support systems must recognize the unique problems caused by representing and reasoning with temporal concepts. This tutorial has three goals: (1) to describe a set of fundamental issues in creating and reasoning with computer models of a changing clinical environment, (2) to present a taxonomy for characterizing the temporal characteristics of computer models of temporal reasoning, and (3) to use this taxonomy to compare the models of time used in implemented medical decision-support programs. Intended audience Those interested in incorporating an ability to reason about events, trends or intervals in medical decision support systems. Some basic understanding of the common reasoning methods will be assumed. Tutor Michael Kahn is assistant professor and chief of the Division of Medical Informatics in the Department of Medicine, Washington University School of Medicine, St Louis and is a practising physician. He has published widely in the area of temporal reasoning and has been associated with a number of influential projects including ONCOCIN. He has wide experience in teaching and lecturing about the subject. Conference Chairman Steen Andreassen, Aalborg Local Chairman Rolf Engelbrecht, Munich Tutorials Chairman Jeremy Wyatt, London International Programme Committee Pedro Barahona, Lisboa Rolf Engelbrecht, Munich Mario Fieschi, Marseille Torgny Groth, Uppsala Roman Hovorka, London Peter Hucklenbroich, Munich Johan van der Lei, Rotterdam Peter McNair, Copenhagen Vassilis Moustakis, Heraklion Pirkko Nykanen, Tampere Kristian Olesen, Aalborg Rory O'Moore, Dublin Francesco del Pozo, Madrid Michael M. Richter, Kaiserslautern Mario Stefanelli, Pavia Jan Talmon, Maastricht Thomas Wetter, Heidelberg Ove Wigertz, Linkoping Jeremy Wyatt, London Pieter Zanstra, Nijmegen Local Organizing Committee Volker Boehm Thomas Diedrich Karl-Heinz Englmeier Norbert Gerzer Lothar Gierl Claudia Hildebrand Lydia Janke Elke Kuehnel Wolfgang Moser Manfred Schnabel Thomas Waschulzik Conference Secretariat AIME 93 c/o GSF Research Center Conference Service Postfach 1129 D - 85758 Oberschleissheim Fax: +49-89/3187-3362