Frank Puppe

SD-map: a fast algorithm for exhaustive subgroup discovery

In this paper we present the novel SD-Map algorithm for exhaustive but efficient subgroup discovery. SD-Map guarantees to identify all interesting subgroup patterns contained in a data set, in contrast to heuristic or sampling-based methods. The SD-Map algorithm utilizes the well-known FP-growth method for mining association rules with adaptations for the subgroup discovery task. We show how SD-Map can handle missing values, and provide an experimental evaluation of the performance of the algorithm using synthetic data.

EuraHS: the development of an international online platform for registration and outcome measurement of ventral abdominal wall hernia repair

BackgroundAlthough the repair of ventral abdominal wall hernias is one of the most commonly performed operations, many aspects of their treatment are still under debate or poorly studied. In addition, there is a lack of good definitions and classifications that make the evaluation of studies and meta-analyses in this field of surgery difficult.Materials and methodsUnder the auspices of the board of the European Hernia Society and following the previously published classifications on inguinal and on ventral hernias, a working group was formed to create an online platform for registration and outcome measurement of operations for ventral abdominal wall hernias. Development of such a registry involved reaching agreement about clear definitions and classifications on patient variables, surgical procedures and mesh materials used, as well as outcome parameters. The EuraHS working group (European registry for abdominal wall hernias) comprised of a multinational European expert panel with specific interest in abdominal wall hernias. Over five working group meetings, consensus was reached on definitions for the data to be recorded in the registry.ResultsA set of well-described definitions was made. The previously reported EHS classifications of hernias will be used. Risk factors for recurrences and co-morbidities of patients were listed. A new severity of comorbidity score was defined. Post-operative complications were classified according to existing classifications as described for other fields of surgery. A new 3-dimensional numerical quality-of-life score, EuraHS-QoL score, was defined. An online platform is created based on the definitions and classifications, which can be used by individual surgeons, surgical teams or for multicentre studies. A EuraHS website is constructed with easy access to all the definitions, classifications and results from the database.ConclusionAn online platform for registration and outcome measurement of abdominal wall hernia repairs with clear definitions and classifications is offered to the surgical community. It is hoped that this registry could lead to better evidence-based guidelines for treatment of abdominal wall hernias based on hernia variables, patient variables, available hernia repair materials and techniques.

KnowWE: a Semantic Wiki for knowledge engineering

Recently, Semantic Wikis showed reasonable success as collaboration platforms in the context of social semantic applications. In this paper, we present a novel approach, that interprets the concept of Semantic Wikis as a knowledge engineering environment, that effectively help to build decision-support systems. We introduce the Semantic Wiki KnowWE, that provides the possibility to define and maintain ontologies together with strong problem-solving knowledge. Thus, the wiki can be used to collaboratively build decision-support systems. These enhancements require extensions of the standard Semantic Wiki architecture by a task ontology for problem-solving and an adapted reasoning process. We discuss these extensions in detail, and we describe a case study in the field of medical emergency systems.

Knowledge reuse among diagnostic problem-solving methods in the Shell-Kit D3

While diagnostic problem solving is in principle well understood, building and maintaining systems in large domains cost effectively is an open issue. Various methods have different advantages and disadvantages making their integration attractive. When switching from one method to another in an integrated system, as much knowledge as possible should be reused. We present an inference structure for diagnostic problem solving optimized with respect to knowledge reuse among methods motivated by experience from large knowledge bases built in medical, technical and other diagnostic domains.

A Diagnostic Expert System for Structured Reports, Quality Assessment, and Training of Residents in Sonography

Abstract.Background:The quality of medical reports on diagnostic procedures has a considerable impact on the quality of medical care. Handwritten or otherwise unstructured reports tend to be incomplete, whereas structured questionnaires are of limited flexibility and not considered case-adequate. Thus, medical reports of this kind may promote an incomplete and misleading documentation and, therefore, be problematic with respect to their reliability.Methods:SonoConsult (SC), an expert system for structured and case-adequate documentation of sonographic findings with an additional diagnostic component, was evaluated with respect to user acceptance and suitability for enhancing the quality of reports and supporting sonographic beginners. The expectations and the attitudes of the users toward the program were evaluated by anonymous questionnaires. The documentation of findings and the diagnostic conclusions in 103 free text reports made by experienced examiners were evaluated by subjecting their information to a subsequent input into SC. Free text reports were checked for information that was asked by SC but not mentioned in the reports. In a series of 150 cases, the system diagnoses were blinded during input of findings into SC-questionnaires and the examiners’ diagnostic conclusions were compared with the uncovered SC-diagnoses with respect to forgotten diagnoses.Results:The structured and data-driven acquisition of information by the program was well accepted by the users. However, only a medium interest in the system-delivered diagnoses was noted. The program-generated reports were characterized by a more detailed description of the findings and a higher number of diagnoses in comparison to the unstructured reports before introduction of SC as the only documentation system. When unaware of the system diagnoses, information was entered into the questionnaires, and SC generated some diagnoses which were not mentioned by the examiners in their conclusions. The possibility to inspect the system diagnoses led to an enhancement of the number of diagnoses the examiners mentioned in their conclusions. By contrast, the examiners meant that the influence of the program on their conclusions was minimal or dispensable. Beginners in sonography acknowledged that the program led them to perform a complete examination in an adequate sequence.Conclusion:An expert system for the data-driven, case-adequate information acquisition of abdominal ultrasound examinations may enhance the quality of the reports and, potentially, of the examinations at the same time. In addition, it may help beginners to learn a structured problem- and finding-adequate examination sequence.Zusammenfassung.Hintergrund:Die Qualität von Befundberichten diagnostischer Untersuchungen hat wesentliche Auswirkungen auf die medizinische Versorgung Kranker. Handschriftliche oder sonstige unstrukturierte Berichte tendieren zu Unvollständigkeit, wohingegen strukturierte Erhebungsbögen nur beschränkt flexibel und nicht immer falladäquat sind. Medizinische Befundberichte dieser Art können eine lückenhafte und irreführende Dokumentation begünstigen und hinsichtlich ihrer Verlässlichkeit problematisch sein.Methodik:SonoConsult (SC), ein Expertensystem zur strukturierten falladäquaten Befunderhebung mit einer zusätzlichen diagnostischen Kompetenz in der Abdominalsonographie, wurde bezüglich seiner Akzeptanz und Eignung zur Steigerung der Befundqualität und zur Unterstützung sonographischer Anfänger evaluiert. Die Erwartungen an das Programm und die Einstellung der Nutzer zum Programm wurden durch anonyme Fragebögen erfasst. Die Inhalte von 103 Freitextberichten erfahrener Untersucher wurden in SC eingegeben und auf solche Befunde hin überprüft, die von SC erfragt wurden, aber nicht im Bericht erwähnt waren. In einer Serie von 150 Untersuchungen wurden die Systemdiagnosen während der Befundeingabe verblindet und die Diagnoseangaben der Untersucher mit den für die Evaluation sichtbar gemachten Diagnosen von SC verglichen.Ergebnisse:Die strukturierte und eingabeabhängige Befunderhebung durch das Programm wurde von den Nutzern akzeptiert. Die Systemdiagnosen waren jedoch nur von mittlerem Interesse. Die vom Programm generierten Befundberichte waren durchschnittlich detaillierter und enthielten mehr diagnostische Schlussfolgerungen als die unstrukturierten Befundberichte vor Einführung von SC als einzigem Dokumentationssystem. Auch bei strukturierter Befundeingabe ohne Kenntnis der Systemdiagnosen wurden von SC im Hintergrund Diagnosen generiert, die vom Untersucher nicht erwähnt wurden. Die Möglichkeit der Einsicht in die Systemdiagnosen führte zur Erhöhung der Zahl der von den Untersuchern in ihren Beurteilungen angegebenen Diagnosen. In Diskrepanz zu diesen Ergebnissen hielten die Untersucher jedoch subjektiv den Einfluss des Systems auf die Qualität der Beurteilungen in ihren Befundberichten für gering bis verzichtbar. Anfänger in der Sonographie bemerkten positiv, dass sie durch das Programm zu einer vollständigen Untersuchung in einer angemessenen Reihenfolge angehalten wurden.Schlussfolgerung:Ein Expertensystem zur eingabeabhängigen, falladäquaten Befunderhebung in der abdominellen Sonographie kann die Qualität der Befundberichte und möglicherweise gleichzeitig die der Untersuchungen erhöhen. Es kann zudem Anfängern helfen, eine problem- und falladäquate Untersuchungssequenz zu erlernen.

Inductive Learning for Case-Based Diagnosis with Multiple Faults

We present adapted inductive methods for learningsimilarities, parameter weights and diagnostic profiles for case-based reasoning. All of these methods can be refined incrementally by applyingdif ferent types of background knowledge. Diagnostic profiles are used for extending the conventional CBR to solve cases with multiple faults. The context of our work is to supplement a medical documentation and consultation system by CBR techniques, and we present an evaluation with a real-world case base.

A case-based approach for characterization and analysis of subgroup patterns

Abstract In general, cases capture knowledge and concrete experiences of specific situations. By exploiting case-based knowledge for characterizing a subgroup pattern, additional information about the subgroup objects can be provided. This paper proposes a case-based approach for characterizing and analyzing subgroup patterns: It presents techniques for retrieving characteristic factors and a set of corresponding cases for the inspection and analysis of a specific subgroup pattern. Then, the set of factors and cases are merged into prototypical cases for presentation to the user. Such an alternative view on the subgroup pattern provides important introspective information on the subgroup objects, that is, the cases covered by the subgroup description: Using drill-down techniques, the user can perform a detailed introspection of a subgroup pattern using prototypical pattern cases. Additionally, these enable a convenient retrieval of interesting (meta-)information associated with the respective subgroup objects.

UIMA Ruta: Rapid development of rule-based information extraction applications

Rule-based information extraction is an important approach for processing the increasingly available amount of unstructured data. The manual creation of rule-based applications is a time-consuming and tedious task, which requires qualified knowledge engineers. The costs of this process can be reduced by providing a suitable rule language and extensive tooling support. This paper presents UIMA Ruta, a tool for rule-based information extraction and text processing applications. The system was designed with focus on rapid development. The rule language and its matching paradigm facilitate the quick specification of comprehensible extraction knowledge. They support a compact representation while still providing a high level of expressiveness. These advantages are supplemented by the development environment UIMA Ruta Workbench. It provides, in addition to extensive editing support, essential assistance for explanation of rule execution, introspection, automatic validation, and rule induction. UIMA Ruta is a useful tool for academia and industry due to its open source license. We compare UIMA Ruta to related rule-based systems especially concerning the compactness of the rule representation, the expressiveness, and the provided tooling support. The competitiveness of the runtime performance is shown in relation to a popular and freely-available system. A selection of case studies implemented with UIMA Ruta illustrates the usefulness of the system in real-world scenarios.

KnowWE: community-based knowledge capture with knowledge wikis

This paper presents a collaborative knowledge engineering approach based on the widespread wiki technique.The interface of a standard wiki system is extended to allow for the capture, the maintenance and the use of knowledge systems.

Subgroup mining for interactive knowledge refinement

When knowledge systems are deployed into a real-world application, then the maintenance of the knowledge is a crucial success factor. In the past, some approaches for the automatic refinement of knowledge bases have been proposed. Many only provide limited control during the modification and refinement process, and often assumptions about the correctness of the knowledge base and case base are made. However, such assumptions do not necessarily hold for real-world applications. In this paper, we present a novel interactive approach for the user-guided refinement of knowledge bases. Subgroup mining methods are used to discover local patterns that describe factors potentially causing incorrect behavior of the knowledge system. We provide a case study of the presented approach with a fielded system in the medical domain.