Submitted by Frederico Azevedo (fazevedo@cdts.fiocruz.br) on 2010-11-10T17:01:50Z No. of bitstreams: 1 oliveiravsm.pdf: 1878124 bytes, checksum: 96c6fd44407809d315bca0121b1f16ff (MD5) / Made available in DSpace on 2010-11-10T17:01:50Z (GMT). No. of bitstreams: 1 oliveiravsm.pdf: 1878124 bytes, checksum: 96c6fd44407809d315bca0121b1f16ff (MD5) Previous issue date: 2005 / Este trabalho analisa as possibilidades de tornar interoperáveis bases de dados heterogêneas, de forma que possam ser pesquisadas através de uma única interface. A biblioteca do Instituto Fernandes Figueira, Unidade materno-infantil da FIOCRUZ é referência em tratamento de doenças de alta complexidade. Estes profissionais buscam a Biblioteca com necessidades de informações voltadas para a decisão clínica. As bases de dados mais pesquisadas por estes profissionais são as bases de dados LILACS, MEDLINE e ACERVOS ONLINE FIOCRUZ. Portanto estas bases norteiam este estudo e são analisadas em algumas facetas como: os procedimentos de descrição e as estruturas das bases, buscando semelhanças e divergências nos índices e nos campos de exibição das bases; a representação temática, caracterizada pela utilização dos Tesauros DeCS e MeSH; e tecnologias utilizadas e/ou compatíveis. Paralelamente busca-se apresentar algumas ferramentas tecnológicas atuais de interoperabilidade como os protocolos Z39.50 e o OAI-PMH, os metabuscadores, o conjunto de metadados Dublin Core e o MetaIAH. Através destas análises delineou-se três modelos conceituais para alcançar a interoperabilidade, são eles: Compartilhamento de Esforços, quando tanto a Interface de Consulta (IC) e os Recursos Informacionais (RI) trabalham para garantir a interoperabilidade; Esforço concentrado nos RIs, quando os Recursos Informacionais arcam com todo o esforço para possibilitar a interoperabilidade; e, Esforço concentrado na IC, quando a Interface de consulta se adapta para suprir as divergências de cada base proporcionando a interoperabilidade. Para finalizar comparam-se os modelos e as bases de dados estudadas destacando-se os modelos que se constituem como alternativas de interoperabilidade para a biblioteca do Instituto Fernandes e quais as vantagens e desvantagens de cada uma delas. / This work analyses the options to achieve interoperability among different library databases so they can be searched from a single Web interface. The databases object of this research are the ones used in The Fernandes Figueira Institute Library, which is a branch of FIOCRUZ specialized in Maternal and Child Health. Library users generally look for medical decision support information. The Library users preferred databases are LILACS, MEDLINE and ACERVOS ONLINE FIOCRUZ. All of them use the DeCS and MeSH Thesaurus to subject description. Each of these databases are analyzed in aspects as access points, bibliographic description and display formats, in order to identify common elements. Technologies and standards to achieve interoperability as Z39.50 and OAI-PMH protocols, meta-search engines, Dublin Core Metadata Set BIREME’s metaIaH meta-search engine are also analyzed. From this analysis emerged three conceptual models to achieve interoperability among the three databases, evolving two actors, the search interface – SI, and the information resources (the three databases) - IR: Effort Sharing, between SI and IRs; Concentrated Effort in the RIs; and Concentrated Effort in the SI to achieve interoperability. The three models are compared, outlining the advantages and disadvantages of each one to achieve interoperability, among the databases of interest to Fernandes Figueira Institute Library.

The goal of this master’s thesis is to identify and evaluate data mining algorithms which are commonly implemented in modern Medical Decision Support Systems (MDSS). They are used in various healthcare units all over the world. These institutions store large amounts of medical data. This data may contain relevant medical information hidden in various patterns buried among the records. Within the research several popular MDSS’s are analyzed in order to determine the most common data mining algorithms utilized by them. Three algorithms have been identified: Naïve Bayes, Multilayer Perceptron and C4.5. Prior to the very analyses the algorithms are calibrated. Several testing configurations are tested in order to determine the best setting for the algorithms. Afterwards, an ultimate comparison of the algorithms orders them with respect to their performance. The evaluation is based on a set of performance metrics. The analyses are conducted in WEKA on five UCI medical datasets: breast cancer, hepatitis, heart disease, dermatology disease, diabetes. The analyses have shown that it is very difficult to name a single data mining algorithm to be the most suitable for the medical data. The results gained for the algorithms were very similar. However, the final evaluation of the outcomes allowed singling out the Naïve Bayes to be the best classifier for the given domain. It was followed by the Multilayer Perceptron and the C4.5.

The scale and severity of cancer, specifically breast cancer, remains significantly different across the spectrum of low-income to high-income countries. This study explores women’s beliefs about breast cancer and associated prevention and health-seeking behaviors in a rural area of Uganda. Through a critical medical anthropological perspective, the study examines the social, cultural, and economic factors that shape women’s understanding of cancer, and breast cancer specifically, and that influence their use of biomedical services. Data were collected over a three-month period through 35 in-depth interviews and two focus groups with 10 women older than 18 years in the rural setting of Nakirebe within Mpigi District, and through five interviews with health care personnel from a private and a government health care facility in Mpigi District. Quantitative and Qualitative data were analyzed using SPSS version 23 and MAXQDA 12.0.2, respectively. Findings suggest that women in this rural setting have limited access to screening and incomplete knowledge about breast cancer, and cancer in general, and internalize fears of a cancer diagnosis. No women were diagnosed with any type of cancer at the time of this study. Common attitudes towards cancer from the women include inevitable death, cancer is caused by contact with artificial substances and/or germs, and cancer causes pain, wounds that never heal, and the removal of body parts. Recommendations for improving cancer control and management in rural Uganda through awareness initiatives and community health outreach programs are presented.

No description available.

No description available.

Diese Dissertation umfasst vier Manuskripte zum Thema Risikokommunikation und medizinischen Entscheidungen. Das erste Manuskript diskutiert Unterschiede, Gemeinsamkeiten und die Anwendbarkeit von drei zentralen Ansätzen, die helfen sollen, bessere Entscheidungen zu treffen (Nudging, Social Marketing, Empowerment). Das zweite Manuskript präsentiert Ergebnisse einer Medienanalyse zur Evaluation von Zeitungs- und Internetberichten in Deutschland und Spanien über die HPV-Impfung. Basierend auf vordefinierten Standards für transparente, vollständige und korrekte Risikokommunikation, deckt die Medienanalyse Schwächen in der Berichterstattung auf. Das dritte Manuskript untersucht wie Laien relative Risikoreduktionen bzw. –erhöhungen, ein Standardformt in der Medizin, verstehen. Beide Formate führen Laien und Experten in die Irre und führen zur Überschätzung der tatsächlichen Effekte. Ein diskutierter Ausweg ist die zusätzliche Kommunikation der Basisrate. Die Ergebnisse zeigen, dass das Verständnis von relativen Risikoreduktionen (-erhöhungen) mit Basisrate von dem Präsentationsformat (Prozent- vs. Häufigkeitsformat) und der individuellen Fähigkeit im Zahlenverständnis abhängt. Teilnehmer mit geringem Zahlenverständnis profitierten von der Darstellung in Häufigkeiten; Teilnehmer mit hohem Zahlenverständnis zeigen ein besseres Verständnis unabhängig des Formats. Dennoch—selbst mit Basisrate—missverstehen viele Teilnehmer die Risikoinformation. Das vierte Manuskript untersucht wie Teilnehmer Behandlungen unter Unsicherheit auswählen. Ein Einwand gegen die Kommunikation von Unsicherheit ist die Behauptung, dass Menschen Unsicherheit in Gewinnsituationen vermeiden, in Verlustsituationen dagegen suchen. Die Ergebnisse dieser Studie in Bezug auf die Auswahl von medizinischen Behandlungen konnten diese Annahmen nicht bestätigen. Darüber hinaus wählte die Mehrheit der Teilnehmer die gleiche Behandlung, wenngleich sich die zugrundeliegende Auswahlstrategie unterschied. / This dissertation comprises four manuscripts focusing on health risk communication and medical decision making. The first manuscript discusses differences, commonalities, and the applicability of three major approaches to help patients make better decisions: nudging, social marketing, and empowerment. The second manuscript presents results of an evaluation of media coverage about the HPV vaccine of newspaper and Internet reports in Germany and Spain. Based on predefined standards for transparent, complete, and correct risk communication, the analysis revealed substantial shortcomings in how the media informed the public. The third manuscript centers on a standard format to communicate treatment benefits and harms: relative risk reductions and increases. Such formats have been found to misinform and mislead patients and health professionals. One suggestion is to always include information about baseline risk to reduce misunderstandings. Results show that even when baseline risk was communicated, it depended on the presentation format (percentage vs. frequency) and people’s numeracy skills whether they correctly interpreted the risk reduction (or increase). Low numerates benefited from a frequency format, whereas high numerates performed better independent of the format. Yet, a substantial proportion of participants still misunderstood the meaning of a relative risk reduction (or increase). The fourth manuscript investigated how laypeople choose between medical treatments when ambiguity is present. One objection against communicating ambiguity is the claim that laypeople are ambiguity averse in the domain of gains and ambiguity seeking in the domain of losses. Results did not find supporting evidence for this claim in medical treatment choice. Moreover, most participants selected the same treatment option, independent of numeracy. However, the underlying choice strategies varied between individuals.

Pharmakodiagnostische Tests eröffnen die Möglichkeit, Krebstherapien individueller auf den Patienten zugeschnitten zu verschreiben. Die vorliegende Dissertation widmet sich deshalb der Frage, wie diese Gruppen in Deutschland sowie den USA in Bezug auf diese Tests Entscheidungen treffen. Alle im Rahmen dieser Arbeit durchgeführten Studien waren unterteilt in eine Vorstudie und eine Hauptsstudie. Die Ergebnisse der Vorstudie wurden im Rahmen der Hauptstudie zur Entwicklung eines Fall-Vignetten Fragebogens benutzt,um die Verwendung von kompensatorischen und nicht-kompensatorischen Entscheidungsstrategien zu untersuchen. Mit Studie I wurde gezeigt, dass sowohl deutsche als auch amerikanische Onkologen eine hohe Bereitschaft haben, solche Tests anzuwenden. Die entsprechenden Entscheidungen wurden am besten durch ein kompensatorisches Modell (Franklin’s Rule)vorhergesagt. Eine Leitlinien-Empfehlung führte nahezu immer zu einer Test-Entscheidung. Verschiedene Bedingungen machten eine Entscheidung für nicht-empfohlene Tests jedoch wahrscheinlicher. Studie II zeigte, dass Pathologen nur zu einem beschränkten Ausmaß bereit waren, von dem etablierten Test-Standard für neuartige Test-Prozeduren abzuweichen. Die Entscheidungsstrategie beider Gruppen wurde gleich gut durch die jeweiligen kompensatorischen Modelle (Franklin’s und Dawes’ Rule) sowie durch das nicht-kompensatorische Modell (Take The Best) vorhergesagt. Für die mit Studie III untersuchten Krebspatienten zeigte sich, dass ein nicht-kompensatorisches Modell (Matching Heuristic) die besten Entscheidungs-Vorhersagen machte.Während die Entscheidungen der US Patienten jedoch maßgeblich von einer Arzt-Empfehlung geleitet waren, fand sich dies nicht für die deutschen Patienten. Die sich aus den Befunden ergebenden Implikationen für die hier untersuchten Gruppen, für die mit der Leitlinien-Entwicklung beauftragten Autoritäten als auch für das Gesundheitssystem im Allgemeinen wurden abschließend diskutiert. / Upcoming pharmacodiagnostic tests offer the opportunity to better tailor cancer treatment decisions to individual patient needs. However, they put oncologists, pathologists, and cancer patients in the position of having to deal with a new technology, which often comes with its own specific risks. Little is known about how these different groups will handle this situation. This thesis is a first effort to examine, within Germany and the USA, how the respective groups would deal with a decision on applying such a test to a cancer treatment decision. All accomplished studies were divided into an explorative pilot study and a main study. Results of the pilot study were used for the main study to develop a case vignette questionnaire in order to investigate compensatory and noncompensatory decision-making strategies.In Study I, it was found that both, German and US oncologists’ decision-making policies were best described by a compensatory model (Franklin’s rule). A recommendation of a test by guidelines triggered nearly always a choice for having the test, although under different conditions also choices for nonrecommended tests were likely. Study II found that pathologists were, to a rather small extent, prepared to opt for more sophisticated test alternatives, compared to standard procedures. For both samples, decision making was equally well-predicted by two compensatory models (Franklin’s rule and Dawes’ rule), as it was by a noncompensatory model (Take The Best.Study III focused on cancer patients. The German as well as the US patients’ decisions were best predicted by a noncompensatory model (Matching Heuristic), while for the US patients, the most impacting cue was the recommendation by an oncologist, what could not be found for the German sample.Several implications of these findings for the respective groups, for authorities in charge of developing guidelines, as well as for the health systems in general, are discussed.

Η τεχνητή νοημοσύνη στη βιοπληροφορική θεωρείται ένα πολύ σημαντικό βήμα αναφορικά με την κατηγοριοποίηση των ασθενειών, ακόμα και τη θεραπεία αυτών. Μέσω των νευρωνικών δικτύων τεχνητής νοημοσύνης μπορούμε να επεξεργαστούμε ιατρική πληροφορία και να κατηγοριοποιήσουμε μοτίβα καίριας σημασίας όσον αφορά την ιατρική διάγνωση. Βέβαια, καθώς στη λήψη αποφάσεων πάντα εισχωρεί ο παράγοντας της αβεβαιότητας, μία από τις πιο κατάλληλες προσεγγίσεις, η οποία προσομοιώνει τον τρόπο που κάθε άνθρωπος λαμβάνει αποφάσεις, είναι η ασαφής λογική. Συνδιάζοντας την ασαφή λογική με τη γνώση ειδικών μπορούμε να μοντελοποιήσουμε σύνθετα φαινόμενα και να αποφανθούμε για τη φύση αυτών. Σε αυτή τη διπλωματική εργασία υλοποιείται ένα ασαφές έξυπνο σύστημα που έχει ως σκοπό να μοντελοποιήσει πέντα καρδιολογικής φύσεως ασθένειες, χρησιμοποιώντας υλικό το οποίο προέρχεται από τη γνώση ειδικών στον τομέα της καρδιολογίας: στεφανιαία νόσος, υπέρταση, κολπική μαρμαρυγή, καρδιακή ανεπάρκεια, διαβήτης. Επιπλέον, το σύστημα, σε συνεργασία με το αρμόδιο ιατρικό προσωπικό, παραμετροποιήθηκε και στη συνέχεια έγινε προσπάθεια βελτιστοποίησής του μέσω της ενσωμάτωσης νευρωνικών δικτύων. Η αποδοτικότητά του αξιολογήθηκε ευνοϊκά μέσα από μία ομάδα ιατρών, δίνοντας ελπίδες για μία νέα εποχή στον τρόπο διεξαγωγής ιατρικής διάγνωσης. Το συγκεκριμένο σύστημα θα αποτελέσει τμήμα του Cardiosmart365, ενός ολοκληρωμένου συστήματος για τη δια βίου παρακολούθηση ασθενών με καρδιολογικά προβλήματα, την έγκαιρη διάγνωση και τη βέλτιστη διαχείριση περιπτώσεων εκτάκτου ανάγκης. Σε αυτό το έξυπνο ασαφές σύστημα προσαρτάται η γνώση που προκύπτει μέσα από τα νευρωνικά δίκτυα, με την οποία και επιτυγχάνεται αυτόματα η βελτιστοποίησή του. / Arti cial intelligence (AI) in bioinformatics is considered to be a great step towards disease classi cation, or even disease treatment. AI gives the opportunity through arti cial neural networks (ANNs) to process medical information and classify pat- terns, something of great importance, as far as medical diagnosis is conserned. How- ever, since there is always the factor of uncertainty in decision making, fuzzy logic is considered to be one of the most suitable approximations, since it deals with reason- ing that is approximate rather than xed and exact, thus closer to human reasoning. Therefore, based on human expert knowledge they are capable of modeling complex phenomena. In this diploma thesis, we implement a fuzzy expert system, consisting of ve subsystems, concerning ve cariological diseases, incorporating expert knowledge on this particular eld: coronary artery disease, hypertension, atrial brillation, heart failure, and diabetes. Moreover, the parameters were con gured, in cooperation with experts on the eld, and optimization e orts were made through the integration of neural networks. Evaluated by a group of doctors, the e ciency was rated as satisfactory, giving hope for a new era in the way medical diagnosis is conducted. This system will be a part of Cardiosmart365, an integrated system for lifelong cardiologic patient monitoring, early detection of emergency, and optimal process management of the emergency incident. In the fuzzy expert system implemented, knowledge through neural networks is incorporated, thus achieving automatic opti- mization.

Cette thèse étudie la distorsion de probabilité dans le jugement clinique afin de comparer le jugement des médecins à des modèles statistiques. Nous supposons que les médecins forment leur jugement clinique en intégrant une composante analytique et une composante intuitive. Dans ce cadre, les médecins peuvent souffrir de plusieurs biais dans la façon dont ils évaluent et intègrent les deux composantes. Cette thèse rassemble les résultats obtenus sur le terrain et en laboratoire. À partir de données médicales, nous avons constaté que les médecins n'étaient pas aussi bons que les modèles statistiques à intégrer des évidences médicales. Ils surestimaient les petites probabilités que le patient soit malade et sous­-estimaient les probabilités élevées. Nous avons constaté que leur jugement biaisé pourrait entraîner un sur­-traitement. Comment améliorer leur jugement? Premièrement, nous avons envisagé de remplacer le jugement du médecin par la probabilité de notre modèle statistique. Pour améliorer la décision, il était nécessaire d'élaborer un score statistique qui combine le modèle analytique, la composante intuitive du médecin et sa déviation observée par rapport à la décision attendue. Deuxièmement, nous avons testé en laboratoire des facteurs qui peuvent influencer le traitement de l'information. Nous avons trouvé que la capacité des participants à apprendre la valeur de la composante analytique, sans feedback externe, dépend de la qualité de leur composante intuitive et de leur mémoire de travail. Nous avons aussi trouvé que la capacité des participants à intégrer les deux composantes dépend de leur mémoire de travail, mais pas de leur évaluation de la composante intuitive. / This thesis studies probability distortion in clinical judgment to compare physicians’ judgment with statistical models. We considered that physicians form their clinical judgment by integrating an analytical component and an intuitive component. We documented that physicians may suffer from several biases in the way they evaluate and integrate the two components. This dissertation gathers findings from the field and the lab. With actual medical data practice, we found that physicians were not as good as the statistical models at integrating consistently medical evidence. They over­estimated small probabilities that the patient had the disease and under­ estimated large probabilities. We found that their biased probability judgment might cause unnecessary health care treatment. How then can we improve physician judgment? First, we considered to replace physician judgment by the probability generated from our statistical model. To actually improve decision it was necessary to develop a statistical score that combines the analytical model, the intuitive component of the physician and his observed deviation from the expected decision. Second, we tested in the lab factors that may affect information processing. We found that participants’ ability to learn about the value of the analytical component, without external feedback, depends on the quality of their intuitive component and their working memory. We also found that participants’ ability to integrate both components together depends on their working memory but not their evaluation of the intuitive component.

Made available in DSpace on 2015-05-14T12:36:36Z (GMT). No. of bitstreams: 1 ArquivoTotalRenata.pdf: 2638998 bytes, checksum: c2e2373b63cd157ac4705018a4e58468 (MD5) Previous issue date: 2013-02-19 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / In recent years, there was an increasing interest in the use of information technology in the medical field. In this way, several studies have been conducted regarding a large range of diseases, including cancer. For example, there are organizations that maintain databases, which record information about cases of cancer around the world, so that health personnel can investigate such bases and find possible classifications for an initial diagnostic according to the symptoms presented by their patients. However, these databases do not offer a proper support for this investigation. This dissertation discusses the use of the of case-based reasoning and rule-based reasoning technologies as a solution to support medical diagnosis via representations of actual patients and adaptations of these cases to define more specialized diagnoses, according to the peculiarities of each patient. The data collection was carried out in the Napoleao Laureano Hospital, at Joao Pessoa city, and the information inherent in the structure of the case, the rules and weights were defined in accordance with the specialized literature and conversations with health professionals. Despite the absence of a specialist, the small base of cases and presence of limited information on each of them, the results showed that the system is effective as a RBC system aimed at diagnosis about cancer. The focus of this project is on the main gastrointestinal cancer domain, but the ideas can be extended to other fields of cancer. / O tratamento de pacientes com câncer é um desafio para os hospitais e centros de saúde. O primeiro problema é classificar ou identificar o tipo específico de câncer. Em seguida, os médicos devem determinar um tratamento adequado para tal doença. Existem bancos de dados mundiais que registram informações sobre câncer, de modo que os médicos possam tentar achar uma classificação para a doença de acordo com os sintomas apresentados pelo paciente. Contudo eles não oferecem um bom suporte para a procura das doenças e, principalmente, para a identificação do tratamento. Este artigo propõe a utilização das tecnologias de raciocínio baseado em casos e raciocínio baseado em regras como solução para suportar estas duas atividades via representações de casos de pacientes e adaptações destes casos para a definição de diagnósticos mais especializados, de acordo com as peculiaridades de cada paciente. Nosso foco é sobre o câncer gastrointestinal, mas as idéias podem ser estendidas para outros domínios de câncer.