No description available.

No description available.

The Preop medical expert system (Langton et. al, 1990) was originally developed using the expert system tool, Nexpert Object, on a VAX computer. Nexpert Object creates an expert system specification which is executed by an interpreter within Nexpert Object. The original implementation, however, has several limitations, including: 1. lack of physical portability 2. requires Nexpert Object to run 3. crude user interface. In order to overcome the first limitation, Preop is implemented on a PC DOS portable computer. This project is addresses the other two limitations. Creating a compiled version of Preop eliminates the need for the Nexpert Object interpreter, and implementing it as a Microsoft Windows application provides a better user interface. / Thesis / Master of Science (MSc)

International Telemetering Conference Proceedings / October 29-November 02, 1990 / Riviera Hotel and Convention Center, Las Vegas, Nevada / Expert systems can be of great benefit to analysts, mission controllers, and flight safety officers who must make real-time decisions based upon a myriad of quickly changing parameters. Operators may require extensive training to view a limited parameter set or they can be aided by computers systematically monitoring many parameters. But, even today’s powerful general purpose computers and workstations are limited in their capabilities to monitor (i.e., process) the desired number of parameters. Thus, applications are distributed across multiple platforms. A telemetry ground station front end system (i.e., a real-time data server) that distributes preprocessed data to multiple knowledge stations over standard communications networks is presented.

This dissertation develops methods to minimize recommendation error costs when inputs to a rule-based expert system are prone to errors. The problem often arises in web-based applications where data are inherently noisy or provided by users who perceive some benefit from falsifying inputs. Prior studies proposed methods that attempted to minimize the probability of recommendation error, but did not take into account the relative costs of different types of errors. In situations where these differences are significant, an approach that minimizes the expected misclassification error costs has advantages over extant methods that ignore these costs. Building on the existing literature, two new techniques - Cost-Based Input Modification (CBIM) and Cost-Based Knowledge-Base Modification (CBKM) were developed and evaluated. Each method takes as inputs (1) the joint probability distribution of a set of rules, (2) the distortion matrix for input noise as characterized by the probability distribution of the observed input vectors conditioned on their true values, and (3) the misclassification cost for each type of recommendation error. Under CBIM, for any observed input vector v, the recommendation is based on a modified input vector v' such that the expected error costs are minimized. Under CBKM the rule base itself is modified to minimize the expected cost of error. The proposed methods were investigated as follows: as a control, in the special case where the costs associated with different types of errors are identical, the recommendations under these methods were compared for consistency with those obtained under extant methods. Next, the relative advantages of CBIM and CBKM were compared as (1) the noise level changed, and (2) the structure of the cost matrix varied. As expected, CBKM and CBIM outperformed the extant Knowledge Base Modification (KM) and Input Modification (IM) methods over a wide range of input distortion and cost matrices, with some restrictions. Under the control, with constant misclassification costs, the new methods performed equally with the extant methods. As misclassification costs increased, CBKM outperformed KM and CBIM outperformed IM. Using different cost matrices to increase misclassification cost asymmetry and order, CBKM and CBIM performance increased. At very low distortion levels, CBKM and CBIM underperformed as error probability became more significant in each method's estimation. Additionally, CBKM outperformed CBIM over a wide range of input distortion as its technique of modifying an original knowledge base outperformed the technique of modifying inputs to an unmodified decision tree.

Active fire protection systems are an essential fire safety management tool, particularly in potentially high financial and risk consequence scenarios. In the UK and Europe over recent decades regulatory changes have been successful in creating an environment in which more innovation can take place. Increased numbers of fixed firefighting system types are now available to the user. However, not all systems offered are equal in terms of; suitability, cost, maturity of supporting knowledge, and overall performance or in-service reliability. Understanding of the systems performance and its limitations and how to match this to the assessed fire risk is incomplete among users. Experts are observing increasing numbers of what they consider to be poor fixed firefighting system choices leading to weaker fire safety designs, which is a cause of concern. Therefore the research aim is to verify that these concerns are founded and, that being the case, to develop a decision support system and related supporting resources to further this aspect of fire safety education and enable users to make better informed system selections. Thus, the focus of this research has been to develop a fixed firefighting system selection tool to complement existing legislation, which incorporates logic, rules and fire safety educational resources in a variety of formats to aid the fire safety design process. A variety of largely heuristic techniques have been used to aggregate data to form knowledge to underpin fixed firefighting system selection tool. In this form, the tool has been validated by experts as being a useful resource. The developed tool also provides ample opportunity for useful ongoing future development. The work recognises that cost and benefit are critical in the selection process. Supporting resources have been incorporated into the tool to assist users in evaluating the levels of reliability they might expect from a system in their circumstances. This tool has now been exposed to a wider audience of experts as part of an evaluation process. Findings include: that the tool is an innovative approach to promoting good fire safety designs, the tool efficiently provides useful fire safety education to users and the developed supporting resources which consider firefighting system reliability are helpful. This thesis and reference papers summarise the key stages of this research and tool development. The thesis concludes by outlining the progress achieved by this work and recommendations arising.

Philosophiae Doctor - PhD / Background: Extended-spectrum beta-lactamase-producing (ESBL) and carbapenemaseproducing Gram-negative bacilli showing resistance to cephalosporins and carbapenems respectively, have been reported from several countries globally and recently among Libyan combatants who have been transferred to European countries for advanced medical care. However, there is a lack of data about their presence in Misrata and in Libya in general. This is the first documented study aimed at investigating the prevalence and resistance mechanisms of ESBL and carbapenemase-producing K. pneumoniae isolates from Misrata. Materials and Methods: Two hundred Gram-negative bacillus isolates were collected and identified from hospitals and pathology laboratories in Misrata. Following antimicrobial susceptibility screening, those showing resistance to cephalosporin and carbapenem were tested for ESBL activity using the Modified double disc synergy test, Sensititer ESBL confirmatory MIC plates and MAST AmpC detection sets D52C and D68C. Carbapenemase activity was detected using RAPIDEC CARBA NP test, Modified Hodge test (MHT), carbapenem inactivation methods (CIM), carbapenem combined test (CCT), and by MAST carba puls set. ESBL and carbapenemases genes were detected using multiplex PCR. Results: K. pneumoniae was the predominant species (85/200) of the 14 species identified, with 56 (65.8%) showing carbapenem resistance, 16 (18.8%) were cephalosporin-resistant carbapenem-susceptible and 13 (15.2%) were susceptible to all antibiotics except ampicillin. OXA-48 was the only carbapenemase detected, with SHV, TEM and CTX-M group 1 found in almost all carbapenem and cephalosporin resistant K. pneumoniae. Rep-PCR analysis revealed multiple clones and some K. pneumoniae strains were genetically related or indistinguishable despite differences in ESBL genes or carbapenemase activity. Conclusion: The findings of this study show that carbapenemase- and ESBL-producing K. pneumoniae are prevalent in Misrata and emphasize the urgent need for optimized infection control and antibiotic stewardship programmes in the Libyan hospitals to prevent further spread of these organisms.

This objectivist, experimental study investigated the influence of virtual expert systems (VES) on the decision accuracy of non-expert consultants within a technology consulting contact centre environment. Because of the overwhelming availability of conceptual information, non-expert consultants experience challenges in making accurate decisions, and would benefit from augmented technologies, such as VES. VES hold the ability to capture and scale large volumes of decision variables for consideration by human experts when making decisions. A total of 40 participants were randomly selected from contact centres in the Eastern Cape and Gauteng provinces of South Africa for this study. Human logic was captured and scaled into a technology fault finding virtual expert and administered as an experiment to group participants. The experimental and control group participants were randomly assigned to the respective groups of 20 participants each. The control group was exposed to the paper-based, fault-finding manual. The pre-test and post-test data were collected based on four decision accuracy measures, namely individual performance, average call handling time, first call resolution and customer service. The Clarify performance system of the participating technology consulting company was used as data collection tool to record the findings used for Chapter 5. Statistical data analyses were performed using ANOVA and two-tailed significance tests to test the relationship between VES and decision accuracy in the pre-test and post-test phases of the study. The study found that the participant scores on the decision accuracy measures were only statistically significant on the measure of first call resolution measure (significance score of a p value <.05). On the other (three) measures, the scores obtained from experimental group participants showed more improvement than that of the control group participants. Consequently, the hypothesis that the use of VES enhance decision accuracy amongst non-expert technology consultants was accepted and the alternative hypothesis rejected. Some limitations pertaining to the resultant Hawthorne effect (the effect when some employees work harder and perform better when they are participants in an experiment) was noted amongst participants. This effect resulted from the use of team leaders in monitoring performance during the experiment and the involvement of the technology consulting company in determining the performance norms of the identified measures. Another limitation of the study related to the size of the sample where only two provinces were included. The limitation may affect the generalisation of results to other future settings when such a study is repeated. It was recommended that future studies in this field should make provision for a larger population, inclusive of other provinces to avoid these limitations.

Many learning disabled student being served by the public school systems have been inaccurately classified. Training and research efforts are needed to assist members of the multidisciplinary team in making more accurate learning disabilities classification decisions. CLASS.LD2, a computer-based expert system, was designed to assist multidisciplinary teams by providing second-opinion advice regarding the appropriateness of a learning disabilities classification for individual student cases. The existing expert system, CLASS.LD2, was combined with strategies for effective concept instruction to create an instructional package entitled LO.Trainer. The purpose of this study was (a) to develop a computer-based instructional package combining expert system technology and strategies for effective concept instruction and (b) to test the effectiveness of the instructional package against another system application. The training application against which the instructional package was compared consisted of users running consultations with the original expert system. Of specific interest was (a) the effectiveness of both training programs across experienced and inexperienced teachers, (b) the performance of the experienced as compared with the inexperienced teachers regardless of the training program used, (c) whether an interaction between level of experience and training program occurred, ( d) which training program was more effective for the experienced teachers, and (e) which training program was more effective for the inexperienced teachers. Ninety-seven students from three universities served as subjects and were randomly assigned to one of the two treatment groups. Subjects who completed the LO.Trainer materials scored statistically (p < .05) and educationally higher (SMD = + 0.96) on the posttest than those who ran CLASS.LD2 consultations. Statistical and educational significance were al so obtained across the experienced and inexperienced subjects when considered alone. An interaction, although not statistically significant (p < .05), was obtained between group and experience level. Although there exist many similarities between the processes of building expert systems and concept analysis, incorporating both to develop an effective training tool had not previously been demonstrated. Results of this study indicated that the two fields, successfully combined, can create an effective and efficient training tool.

An expert system for identifying cows to be culled, MAXCULL, was programmed to run on an IBM or compatible personal computer . It was designed to be used with Dairy Herd Improvement (DHI) records as an aid in decision making. MAXCULL used fifty-two health, seventeen reproduction, and thirty-nine production rules in the analysis. MAXCULL was initially developed using two expert system tools. VP-Expert and Super Expert both had inductive abilities and were reasonably priced . VP - Expert was selected to continue the development of the MAXCULL system . The program u sed a rule-based method of storing knowledge, which was obtained from literature reviewed in the health, reproduction and management areas. Three blocks of rules were developed. MAXCULL used a backward- chaining control strategy. The information on each cow was obtained from a special report obtained from DHI Provo . The diagnosis from MAXCULL produced an explanation paragraph identifying possible reasons for removing the cow. Twenty herds with DHI records were identified, ten assigned as controls and ten to be evaluated with the MAXCULL system . General linear model procedures were used to compare thirteen variables after using MAXCULL for one year. No significant differences were noted for any of the variables. The chi- square analysis showed that the decisions of MAXCULL were significantly different from the decisions of the manager. The final survey supports the idea that dairy management expertise can be provided to the dairy manager through an expert system.