Biomedical subjects

F Alemi

Publications and source records attributed to F Alemi.

40 records · Page 3Linked to original sources

PubMedDate not supplied

Continuous improvement evaluation: a framework for multisite evaluation studies.

This article proposes the use of the continuous improvement evaluation (CIE), a framework for multisite demonstration or evaluation studies. This framework is designed for studying intervention programs that change during the evaluation. The development of family drug courts is provided as an example. CIE relies on outcome data collected over time and benchmarked across similar cases in comparison sites; thus, this study was designed to collect data on effectiveness of intervention programs at multiple sites and over time. A weight is calculated for similarity of any two cases based on features they share. In statistical process control charts, these weights are used to compare outcomes at the site against the average of similar cases in comparison groups. Once data are benchmarked, program staff meet to discuss process changes that have led to improvements in outcomes. To ensure that intervention programs have access to evaluation reports on demand, information technology is used to collect, clean, and pool data. Computers generate study reports, and evaluators review reports after release to clients. Statistical tools can be used to evaluate changing programs. Traditional evaluators may be concerned about some threats to validity associated with CIE. The article concludes with a discussion of typical threats to validity and how these threats are addressed in the CIE framework.

Data Collection↗

PubMedDate not supplied

An evaluation of factors influencing Bayesian learning systems.

OBJECTIVES: To examine the influences of situational and model factors on the accuracy of Bayesian learning systems. DESIGN: This study examines the impacts of variations in two situational factors, training sample size and number of attributes, and in two model factors, choice of Bayesian model and criteria for excluding model attributes, on the overall accuracy of Bayesian learning systems. MEASUREMENTS: The test data were derived from myocardial infarction patients who were admitted to eight hospitals in New Orleans during 1985. The test sample consisted of 339 cases; the training samples included 100, 400, and 800 cases. APACHE II variables were used for the model attributes and patient discharge status as the outcome predicted. Attribute sets were selected in sizes of 4, 8, and 12. The authors varied the Bayesian models (proper and simple) and the attribute exclusion criteria (optimism and pessimism). RESULTS: The simple Bayes model, which assumes conditional independence, consistently equalled or outperformed the proper (maximally dependent) Bayes model, which assumes conditional dependence, across all training sample and attribute set sizes. Not excluding model attributes was found to be preferable to using sample theory as an attribute exclusion criterion in both the simple and the proper models. CONCLUSION: In the domain tested, the simple Bayes model with optimistic exclusion is more robust than previously assumed and increasing the number of attributes in a model had a greater relative impact on model accuracy than did increasing the number of training sample cases. Assessment of applicability of these findings to other domains will require further study. In addition, other models that are between these two extremes must be investigated. These include models that approximate proper Bayes' conditional dependence computations while requiring fewer training sample cases, attribute exclusion criteria between optimism and pessimism that improve accuracy, and ordering techniques for introducing attributes into Bayes models that optimize the information value associated with the attributes in test-sample cases.

APACHE↗

PubMedDate not supplied

Improving the accuracy of severity indexes: exceptions to the rules.

This study explores the difference between additive and non-additive indexes in measuring the severity of myocardial infarction. It shows, as an example, the fallacy of adding severity scores in a straightforward manner. An additive severity index was constructed from the judgments of seven experts. The experts also identified several exceptions to the additive index. The study used the exceptions to modify the additive index and produce a non-additive severity index. The non-additive severity index explained 36% more of the variance in the severity judgments made by five physicians and two nurses on 50 hypothetical cases than the additive index did. In addition, the non-additive index was 3% more accurate in predicting in-hospital mortality of 7,500 patients with myocardial infarction. When the study reduced the noise in the data by ignoring 1,200 rare cases in which stable estimates of mortality rate were unavailable, the prediction of the non-additive index was 13% more accurate than that of the additive index. Statistical tests showed that the differences between the additive and the non-additive indexes were significant at an alpha level below 1%. The practical implications of non-additive severity indexes are discussed. Researchers and physicians who assess the severity of myocardial infarction should systematically explore exceptions that may improve the accuracy of prediction of an additive index.

Evaluation Studies as Topic↗

PubMedDate not supplied

Automated monitoring of outcomes: application to treatment of drug abuse.

This paper suggests a new approach for lowering follow-up costs, improving the delivery of health care, and monitoring treatment outcomes. An automated telephone follow-up system that calls, identifies, and interviews clients is an alternative method for monitoring patients that may be both reliable and cost-effective. To test the viability of such a system, the authors monitored a patient population that has historically been shown to be difficult to follow: recovering drug users and alcoholics. Forty-two subjects were asked to call the computer and complete interviews on a weekly basis for five months. Clients answered 25 recorded questions by pressing the keys on their telephone pads. The computer automatically analyzed the clients' answers and estimated a probability of relapse for each client. In addition, the computer automatically called subjects who failed to complete interviews at the scheduled times. The study showed that self-reported data collected by a computer are as reliable as data obtained through a written questionnaire and that clients are more willing to respond to computer interviews than to mailed written questionnaires. This study also provides preliminary data on the predictive ability of a questionnaire for predicting relapse.

Adult↗