Diss. (sammanfattning) Uppsala : Univ., 2001. / Härtill 5 uppsatser.

Thesis (Ph. D.)--University of Washington, 2000. / Vita. Includes bibliographical references (leaves 131-133).

Source: Masters Abstracts International, Volume: 47-03, page: 1686. Advisers: Jack J. Lee; Yunxin Fu. Includes bibliographical references.

Purpose: To investigate reasons for patients not proceeding to resective epilepsy surgery after subdural grid evaluation (SDE). To correlate noninvasive investigation results with invasive EEG observations in a set of patients with nonlesional brain MRIs. Methods: Retrospective study of adult epilepsy patients undergoing SDE during an 8-year period at Cleveland Clinic. Construction of semiquantitative "scores" and Bayesian predictors summarizing the localizing value and concordance between noninvasive parameters in a subset with nonlesional MRIs. Results: One hundred forty patients underwent SDE, 25 of whom were subsequently denied resective surgery. In 10 of 25, this was caused by a nonlocalizing subdural ictal EEG onset. Eight of 10 such patients were nonlesional on MRI. Among all nonlesional patients (n = 34 of 140), n 1 = 10 of 34 patients had nonlocalizing and n2 = 24 of 34 had localizing, subdural ictal onsets. As groups, n1 and n 2 were statistically disjoint relative to their noninvasive scores. Bayesian measures predictive of focal invasive ictal EEG were highest for complete concordance of noninvasive parameters, decreasing with lesser degrees of concordance. A localizing scalp interictal EEG was a particularly good Bayesian prognosticator. Conclusions: A small but significant proportion of SDE patients are denied subsequent therapeutic resective surgery. This is due to several reasons, including a nonlocalizing intracranial ictal EEG. The majority of such patients have nonlesional MRIs. The noninvasive data may be summarized by a semiquantitative score, as well as Bayesian likelihood ratios, which correlate with subsequent invasive outcome. This approach may find use in the selection and counseling of potential surgical candidates offered SDE.

Yes / Many studies have examined the extent to which individuals’ probability judgments depart from Bayes’ theorem when revising probability estimates in the light of new information. Generally, these studies have not considered the implications of such departures for decisions involving risk. We identify when such departures will occur in two common types of decisions. We then report on two experiments where people were asked to revise their own prior probabilities of a forthcoming economic recession in the light of new information. When the reliability of the new information was independent of the state of nature, people tended to overreact to it if their prior probability was low and underreact if it was high. When it was not independent, they tended to display conservatism. We identify the circumstances where discrepancies in decisions arising from a failure to use Bayes’ theorem were most likely to occur in the decision context we examined. We found that these discrepancies were relatively rare and, typically, were not serious.

The application of Bayesian networks within the field of institutional research is explored through the development of a Bayesian network used to predict first- to second-year retention of undergraduates. A hybrid approach to model development is employed, in which formal elicitation of subject-matter expertise is combined with machine learning in designing model structure and specification of model parameters. Subject-matter experts include two academic advisors at a small, private liberal arts college in the southeast, and the data used in machine learning include six years of historical student-related information (i.e., demographic, admissions, academic, and financial) on 1,438 first-year students. Netica 5.12, a software package designed for constructing Bayesian networks, is used for building and validating the model. Evaluation of the resulting model’s predictive capabilities is examined, as well as analyses of sensitivity, internal validity, and model complexity. Additionally, the utility of using Bayesian networks within institutional research and higher education is discussed. The importance of comprehensive evaluation is highlighted, due to the study’s inclusion of an unbalanced data set. Best practices and experiences with expert elicitation are also noted, including recommendations for use of formal elicitation frameworks and careful consideration of operating definitions. Academic preparation and financial need risk profile are identified as key variables related to retention, and the need for enhanced data collection surrounding such variables is also revealed. For example, the experts emphasize study skills as an important predictor of retention while noting the absence of collection of quantitative data related to measuring students’ study skills. Finally, the importance and value of the model development process is stressed, as stakeholders are required to articulate, define, discuss, and evaluate model components, assumptions, and results.

Thesis (Ph. D.)--University of Washington, 2000. / Vita. Includes bibliographical references (leaves 107-111).

Thesis (Ph. D.)--University of Washington, 2002. / Vita. Includes bibliographical references (p. 1226-131).

Thesis (Ph. D.)--University of Washington, 2002. / Vita. Includes bibliographical references (leaves 102-108).

Disease maps are geographical maps that display local estimates of disease risk. When the disease is rare, crude risk estimates can be highly variable, leading to extreme estimates in areas with low population density. Bayesian hierarchical models are commonly used to stabilize the disease map, making them more easily interpretable. By exploiting assumptions about the correlation structure in space and time, the statistical model stabilizes the map by shrinking unstable, extreme risk estimates to the risks in surrounding areas (local spatial smoothing) or to the risks at contiguous time points (temporal smoothing). Extreme estimates that are based on smaller populations are subject to a greater degree of shrinkage, particularly when the risks in adjacent areas or at contiguous time points do not support the extreme value and are more stable themselves. / A common goal in disease mapping studies is to identify areas of elevated risk. The objective of this thesis is to compare the accuracy of several fully Bayesian hierarchical models in discriminating between high-risk and background-risk areas. These models differ according to the various spatial, temporal and space-time interaction terms that are included in the model, which can greatly affect the smoothing of the risk estimates. This was accomplished with simulations based on the cervical cancer rate of Kentucky and at-risk person-years of the state of Kentucky's 120 counties from 1995 to 2002. High-risk areas were 'planted' in the generated maps that otherwise had background relative risks of one. The various disease mapping models were applied and their accuracy in correctly identifying high- and background-risk areas was compared by means of Receiver Operating Characteristic curve methodology. Using data on Multiple Sclerosis (MS) on the island of Sardinia, Italy we apply the more successful models to identify areas of elevated MS risk.