Hans-Peter Buscher

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

Abstract.Background:The quality of medical reports on diagnostic proceduresnhas a considerable impact on the quality of medical care.nHandwritten or otherwise unstructured reports tend to benincomplete, whereas structured questionnaires are of limitednflexibility and not considered case-adequate. Thus, medicalnreports of this kind may promote an incomplete and misleadingndocumentation and, therefore, be problematic with respect tontheir reliability.Methods:SonoConsult (SC), an expert system for structured andncase-adequate documentation of sonographic findings with annadditional diagnostic component, was evaluated with respect tonuser acceptance and suitability for enhancing the quality ofnreports and supporting sonographic beginners. The expectationsnand the attitudes of the users toward the program were evaluatednby anonymous questionnaires. The documentation of findings andnthe diagnostic conclusions in 103 free text reports made bynexperienced examiners were evaluated by subjecting theirninformation to a subsequent input into SC. Free text reportsnwere checked for information that was asked by SC but notnmentioned in the reports. In a series of 150 cases, the systemndiagnoses were blinded during input of findings intonSC-questionnaires and the examiners’ diagnostic conclusions werencompared with the uncovered SC-diagnoses with respect tonforgotten diagnoses.Results:The structured and data-driven acquisition of informationnby the program was well accepted by the users. However, only anmedium interest in the system-delivered diagnoses was noted. Thenprogram-generated reports were characterized by a more detailedndescription of the findings and a higher number of diagnoses inncomparison to the unstructured reports before introduction of SCnas the only documentation system. When unaware of the systemndiagnoses, information was entered into the questionnaires, andnSC generated some diagnoses which were not mentioned by thenexaminers in their conclusions. The possibility to inspect thensystem diagnoses led to an enhancement of the number ofndiagnoses the examiners mentioned in their conclusions. Byncontrast, the examiners meant that the influence of the programnon their conclusions was minimal or dispensable. Beginners innsonography acknowledged that the program led them to perform ancomplete examination in an adequate sequence.Conclusion:An expert system for the data-driven, case-adequateninformation acquisition of abdominal ultrasound examinations maynenhance the quality of the reports and, potentially, of thenexaminations at the same time. In addition, it may helpnbeginners to learn a structured problem- and finding-adequatenexamination sequence.Zusammenfassung.Hintergrund:Die Qualität von Befundberichten diagnostischernUntersuchungen hat wesentliche Auswirkungen auf die medizinischenVersorgung Kranker. Handschriftliche oder sonstigenunstrukturierte Berichte tendieren zu Unvollständigkeit,nwohingegen strukturierte Erhebungsbögen nur beschränkt flexibelnund nicht immer falladäquat sind. Medizinische Befundberichtendieser Art können eine lückenhafte und irreführendenDokumentation begünstigen und hinsichtlich ihrer Verlässlichkeitnproblematisch sein.Methodik:SonoConsult (SC), ein Expertensystem zur strukturiertennfalladäquaten Befunderhebung mit einer zusätzlichenndiagnostischen Kompetenz in der Abdominalsonographie, wurdenbezüglich seiner Akzeptanz und Eignung zur Steigerung dernBefundqualität und zur Unterstützung sonographischer Anfängernevaluiert. Die Erwartungen an das Programm und die Einstellungnder Nutzer zum Programm wurden durch anonyme Fragebögen erfasst.nDie Inhalte von 103 Freitextberichten erfahrener Untersuchernwurden in SC eingegeben und auf solche Befunde hin überprüft,ndie von SC erfragt wurden, aber nicht im Bericht erwähnt waren.nIn einer Serie von 150 Untersuchungen wurden die Systemdiagnosennwährend der Befundeingabe verblindet und die Diagnoseangaben dernUntersucher mit den für die Evaluation sichtbar gemachtennDiagnosen von SC verglichen.Ergebnisse:Die strukturierte und eingabeabhängige Befunderhebungndurch das Programm wurde von den Nutzern akzeptiert. DienSystemdiagnosen waren jedoch nur von mittlerem Interesse. Dienvom Programm generierten Befundberichte waren durchschnittlichndetaillierter und enthielten mehr diagnostischenSchlussfolgerungen als die unstrukturierten Befundberichte vornEinführung von SC als einzigem Dokumentationssystem. Auch beinstrukturierter Befundeingabe ohne Kenntnis der Systemdiagnosennwurden von SC im Hintergrund Diagnosen generiert, die vomnUntersucher nicht erwähnt wurden. Die Möglichkeit der Einsichtnin die Systemdiagnosen führte zur Erhöhung der Zahl der von dennUntersuchern in ihren Beurteilungen angegebenen Diagnosen. InnDiskrepanz zu diesen Ergebnissen hielten die Untersucher jedochnsubjektiv den Einfluss des Systems auf die Qualität dernBeurteilungen in ihren Befundberichten für gering bisnverzichtbar. Anfänger in der Sonographie bemerkten positiv, dassnsie durch das Programm zu einer vollständigen Untersuchung inneiner angemessenen Reihenfolge angehalten wurden.Schlussfolgerung:Ein Expertensystem zur eingabeabhängigen, falladäquatennBefunderhebung in der abdominellen Sonographie kann die Qualitätnder Befundberichte und möglicherweise gleichzeitig die dernUntersuchungen erhöhen. Es kann zudem Anfängern helfen, einenproblem- und falladäquate Untersuchungssequenz zunerlernen.

HepatoConsult: a knowledge-based second opinion and documentation system

HepatoConsult is a publicly available knowledge-based second opinion and documentation system aiding in the diagnosis of liver diseases. The positive results of a prospective diagnostic evaluation study encouraged its use in clinical routine, although the available hardware infrastructure was not optimal. The comments of the physicians who used the system confirmed the results of the study and showed that the time for data entering is acceptable and the implicit standardization of terminology and documentation is welcome. Suggestions for improvement included the interface to enter data more easily, the scope to be usable for more patients and the additional capability to generate medical reports from the data.

A SEMI-AUTOMATIC APPROACH FOR CONFOUNDING-AWARE SUBGROUP DISCOVERY

This paper presents a semi-automatic approach for confounding-aware subgroup discovery: Confounding essentially disturbs the measured effect of an association between variables due to the influence of other parameters that were not considered. The proposed method is embedded into a general subgroup discovery approach, and provides the means for detecting potentially confounded subgroup patterns, other unconfounded relations, and/or patterns that are affected by effect-modification. Since there is no purely automatic test for confounding, the discovered relations are presented to the user in a semi-automatic approach. Furthermore, we utilize (causal) domain knowledge for improving the results of the algorithm, since confounding is itself a causal concept. The applicability and benefit of the presented technique is illustrated by real-world examples from a case-study in the medical domain.

Clinical experiences with a knowledge-based system in sonography (sonoconsult)

We evaluated the clinical effects of the knowledge-based documentation and diagnosis system SonoConsult for sonography, which has been used in clinical routine for more than 2 years. The evaluation focuses on the following aspects from the clinical point of view: quality of documentation, quality of diagnostic conclusions, training effects, and research effects. In contrast to wide-spread expectations in the knowledge-based community, the diagnostic conclusions were less important than the other aspects, being much more welcomed by clinicians.

Onto Confounding-Aware Subgroup Discovery

This paper presents a semi-automatic approach for confounding-aware subgroup discovery: We present a method that provides the means for detecting potentially confounded subgroup patterns, other unconfounded relations, and/or patterns that are affected by effect- modification. Since there is no purely automatic test for confounding, the discovered relations are subsequently presented to the user in a semi-automatic approach. Furthermore, we show how to utilize (causal) domain knowledge for improving the results of the algorithm, since confounding is itself a causal concept. The applicability and benefit of the presented technique is illustrated by examples from a case-study in the medical domain.