Wai Yat Wong

Decision support in non-conservative domains: generalization with neural networks

Models in conventional decision support systems (DSSs) are best suited for problem solutions in domains with well defined/structured (mathematical) or partially defined/semi-structured (heuristic) domain models. Nonconservative/unstructured domains are those which either lack a known model or have a poorly defined domain model. Neural networks (NNs) represent an alternative modelling technique which can be useful in such domains. NNs autonomously learn the underlying domain model from examples and have the ability to generalize, i.e., use the learnt model to respond correctly to previously unseen inputs. This paper describes three different experiments to explore the use of NNs for providing decision support by generalization in non-conservative/ unstructured domains. Our results indicate that NNs have the potential to provide adequate decision support in non-conservative/unstructured domains.

Quantitative reverse transcriptase-PCR amplification of cytokine mRNA in liver biopsy specimens using a non-competitive method

Reverse transcriptase‐PCR (RT‐PCR) amplification of mRNA is often the only technique able to detect expression of cytokine mRNA In small samples. The aim of this work was to investigate the utility of a non‐competitive RT‐PCR which used external standards to quantitate TNF‐α mRNA in liver biopsy specimens from liver transplant patients. It involved removal of aliquots from the PCR reaction at successive cycles, followed by dot‐blotting of the samples onto nylon membrane and hybridization with a radioactively‐labelled internal probe. Phosphorimage analysis of the labelled membranes allowed quantitation of the relative amount of PCR product at successive cycles. Plots of log(counts) versus cycle number showed straight lines in the exponential phase of amplification. The slopes of these lines showed the efficiency of amplification, which ranged from 76 to 87% for liver biopsy samples. Estimation of liver biopsy levels of TNF‐α in two separate PCR amplifications showed low inter‐assay variability (r2 = 0.98), Comparison of two separate cDNA syntheses also showed good correlation (r2 = 0.81, P < 0.0001), although not as good as for the PCR alone. This shows that variation in efficiency of cDNA synthesis is likely to contribute as much or more to variability of the analysis as variations in PCR amplification.

A knowledge-based framework for clinical incident management

Abstract Clinical incidents, which occur during the provision of health care, can be costly and deadly. Over three-quarter of these incidents is preventable according to the studies in general practice in Australia ( Bhasale, A., Miller, G., Reid, S., & Britt, H., (1998). Analysing potential harm in Australian general practice: an incident-monitoring study. MJA, 169, 73–76 ). It is important that we learn as much as possible from these incidents to prevent them in the future and improve quality of care. This paper introduces a holistic system, which amalgamates case-based reasoning, rule-based reasoning, causal-based reasoning and an ontological knowledge base for managing clinical incidents in general practice. Clinical incident management includes incident analysis, incident case browsing, statistics and explanation. The system enables health professionals to share the medical incident information, which has caused harm and can cause potential harm. The re-use of such information may prevent or mitigate human or medical errors. Such a hybrid approach provides an effective management of adverse clinical incidents for quality improvement in General Practice.

Automatic speech recognition for generalised time based media retrieval and indexing

Perrrrks]onto make dig]ial or hard copies of all or parr of xh~ work for personal or cl-room use is granted uithout fee provided Ihat copies are nol made or distribmed for profi! or commercial advantage, and ~ha~ copies bear ~hls notice and ihe full citation on the first page. To copy o~hewise, to republis~ lo post on servers or to redistribute 10 lists, requires prior specific permission and or a fee.

Towards an Experience Based Negotiation Agent

Current E-Commerce trading agents with electronic negotiation facilities usually use predefined and non-adaptive negotiation mechanisms [5]. This paper presents a negotiation agent that applies Case-Based Reasoning techniques to capture and re-use previously successful negotiation experiences. This Experience Based Negotiation (EBN) agent provides adaptive negotiation strategies that can be generated dynamically and are context-sensitive. We demonstrate this negotiation agent in the context of used-car trading. This paper describes the negotiation process and the conceptual framework of the EBN agent. It discusses our web based used-car trading prototype, the representation of the used-car trading negotiation experience, and the stages of experience based negotiation. The paper also discusses some experimental observation and illustrates an example of adaptive behaviour exhibited by the EBN agent. We believe that this Experience Based Negotiation framework can enhance the negotiation skills and performance of current trading agents.

A knowledge-based approach to design of intervention strategies for incidents

Presents a knowledge-based approach for design of intervention strategies for clinical incidents, which helps medical professionals explore opportunities for care quality improvement. The approach consists of two processes: incident analysis and intervention advice. Given a specific clinical incident description, the proposed approach enables the identification of underlying problems in the incident and the suggestion of appropriate intervention strategies that can be implemented to prevent repetitions of similar incidents. Case-based reasoning (CBR) and rule-based reasoning (RBR) are applied to generate intervention strategies by adapting ones applied to previous incidents and/or using generalised rule-based knowledge for interventions.

Cyberoos'99: Tactical Agents in the RoboCup Simulation League

This paper describes a framework for formalising tactical reasoning in dynamic multi-agent systems, populated by synthetic (software) agents. The proposed framework is based on a hierarchy of synthetic agent architectures and is expressive enough to capture a subset of desirable properties from both the situated automata and subsumption-style architectures, while retaining the rigour and clarity of logic-based possible worlds semantics. This framework is successfully realised in the RoboCup Simulation League domain. Not only did it provide a solid design approach to object-orientation, but it also enabled incremental implementation and testing of software agents and their modules. In particular, the framework allowed us to correlate enhancements in the agent architecture with tangible improvements in team performance. Cyberoos98 was 3rd place winner of the Pacific Rim series at PRICAI-98. Cyberoos99 finished in the top 18 of the RoboCup-99.