Julie Cowie

Quality of data model for supporting mobile decision making

This paper describes research towards implementation of a mobile decision support system. Our view is that the mobile decision maker will benefit if provided with a measure of the Quality of the Data (QoD) used in deriving a decision, and how QoD improves or deteriorates while he/she is on the move. We propose a QoD model taking into account static and dynamic properties of the mobile decision context, and use multicriteria decision analysis to represent decision model and derive a QoD measure. A prototype mobile decision support system has been developed to investigate the usefulness of the proposed QoD model.

The development of a side effect risk assessment tool (ASyMS©-SERAT) for use in patients with breast cancer undergoing adjuvant chemotherapy

Abstract Patients with breast cancer receiving chemotherapy are at risk of developing toxicities which can be severe or life threatening. The aim of this study was to develop and test a side effect risk modeling tool (ASyMS©-SERAT) for use in patients with breast cancer undergoing adjuvant chemotherapy. The study was conducted in two phases. Phase 1 entailed the development of the ASyMS©-SERAT tool using a secondary data set and in collaboration with an expert group of clinicians and an advisory group of patients. In phase 2, the predictive accuracy of the tool was measured using a prospective data set of 24 patients with breast cancer undergoing adjuvant chemotherapy. A high level of accuracy was reported for four of the six symptoms measured (>70%) supporting the future development and application of ASyMS©-SERAT in the prediction of chemotherapy-related toxicity. For patients, such information can be used to target information on side effects that they are likely to experience thereby facilitating the provision of tailored information based on their individual needs. For clinicians, knowing the likelihood of potential side effects can assist them in identifying patients who are at greater risk of developing certain toxicities, facilitating more targeted and cost-effective interventions.

Support for real-time decision making in Mobile Financial Applications

AbstractsMobile users making real-time decisions based on current information need confidence that their context has been taken into consideration in producing the system’s recommendations. This chapter reviews current use of mobile technologies for context-aware real-time decision support. Specifically, it describes a framework for assessing the impact of mobility in decision making. The framework uses dynamic context model of data quality to represent uncertainties in the mobile decision-making environment. This framework can be used for developing visual interactive displays for communicating to the user relevant changes in data quality when working in mobile environments. As an illustration, this chapter proposes a real-time decision support procedure for on-the-spot assistance to the mobile consumer when choosing the best payment option to efficiently manage their budget. The proposed procedure is based on multi-attribute decision analysis, scenario reasoning, and a quality of data framework. The feasibility of the approach is demonstrated with a mobile decision-support system prototype implementation.

A mobile knowledge management and decision support tool for soil analysis

This paper describes the implementation and evaluation of a mobile knowledge management and decision support system to assist archaeologists in dealing with soils. Our view is that provision of a mobile tool which provides access to expert knowledge and a means of recording pertinent onsite information will be of great benefit in ensuring crucial information about an excavation is not lost and that the excavation proceeds in an appropriate manner. In this paper we discuss the tool developed, and detail how it has been evaluated via a variety of workshop sessions with likely users, and discussions with advisory groups.

Directed intervention crossover applied to bio-control scheduling

This paper describes two directed intervention crossover approaches that are applied to a bio-control dynamic system. Unlike traditional uniform crossover, both the calculated expanding bin (CalEB) method and targeted intervention with stochastic selection (TInSSel) approach actively choose an intervention level and spread based on the fitness of the parents selected for crossover. Results indicate that these approaches lead to significant improvements over uniform crossover (UC) when a penalty is introduced for each intervention point used by the crossover algorithm.

Optimisation and fitness modelling of bio-control in mushroom farming using a Markov network eda

In this paper we present an application of an Estimation of Distribution Algorithm (EDA) that uses a Markov network probabilistic model. The application is to the problem of bio-control in mushroom farming, a domain which admits bang-bang-control solutions. The problem is multi-objective and uses a weighted fitness function. Previous work on this problem has applied genetic algorithms (GA) with directed intervention crossover schemes aimed at effective biocontrol at an efficient level of intervention. Here we compare these approaches with the EDA Distribution Estimation Using Markov networks (DEUMd). DEUMd constructs a probabilistic model using Markov networks. Our experiments compare the quality of solutions produced by DEUMd with the GA approaches and also reveal interesting differences in the search dynamics that have implications for algorithm design.

Real-time management of chemotherapy toxicity using the Advanced Symptom Management System (ASyMS).

This paper describes an ongoing study of the Advanced Symptom Management System (ASyMS) for patients receiving chemotherapy for breast or colorectal cancer. We begin by detailing the ASyMS work to date, providing an overview of research conducted in the field over the last ten years. The current study, ASyMS-III, is then presented, highlighting the study methodology, multi-site involvement, the outcomes being measured, and discussion of the tool. The paper concludes with reflections on the progress of the ASyMS-III study to date, and discusses potential directions for future research.

Fitness directed intervention crossover approaches applied to bio-scheduling problems

This paper discusses the effects of using directed intervention crossover approaches with Genetic Algorithms (GA) and demonstrates their application to scheduling of bio-control agents and cancer chemotherapy treatments. Unlike traditional approaches such as Single Point Crossover (SPC) or Uniform Crossover (UC), the directed intervention techniques actively choose the intervention level based on the fitness of the parents selected for crossover. This work shows that a fitness directed intervention crossover approach leads to significant improvements over SPC and UC when applied to the two different scheduling problems.

Development and preliminary psychometric properties of the Care Experience Feedback Improvement Tool (CEFIT)

Objective To develop a structurally valid and reliable, yet brief measure of patient experience of hospital quality of care, the Care Experience Feedback Improvement Tool (CEFIT). Also, to examine aspects of utility of CEFIT. Background Measuring quality improvement at the clinical interface has become a necessary component of healthcare measurement and improvement plans, but the effectiveness of measuring such complexity is dependent on the purpose and utility of the instrument used. Methods CEFIT was designed from a theoretical model, derived from the literature and a content validity index (CVI) procedure. A telephone population surveyed 802 eligible participants (healthcare experience within the previous 12 months) to complete CEFIT. Internal consistency reliability was tested using Cronbachs α. Principal component analysis was conducted to examine the factor structure and determine structural validity. Quality criteria were applied to judge aspects of utility. Results CVI found a statistically significant proportion of agreement between patient and practitioner experts for CEFIT construction. 802 eligible participants answered the CEFIT questions. Cronbachs α coefficient for internal consistency indicated high reliability (0.78). Interitem (question) total correlations (0.28–0.73) were used to establish the final instrument. Principal component analysis identified one factor accounting for 57.3% variance. Quality critique rated CEFIT as fair for content validity, excellent for structural validity, good for cost, poor for acceptability and good for educational impact. Conclusions CEFIT offers a brief yet structurally sound measure of patient experience of quality of care. The briefness of the 5-item instrument arguably offers high utility in practice. Further studies are needed to explore the utility of CEFIT to provide a robust basis for feedback to local clinical teams and drive quality improvement in the provision of care experience for patients. Further development of aspects of utility is also required.

Bio-control in mushroom farming using a Markov network EDA

In this paper we present an application of an Estimation of Distribution Algorithm (EDA) that uses a Markov network probabilistic model. The application is to the problem of bio-control in mushroom farming, a domain which admits bang-bang-control solutions. The problem is multi-objective and uses a weighted fitness function. Previous work on this problem has applied genetic algorithms (GA) with directed intervention crossover schemes aimed at effective biocontrol at an efficient level of intervention. Here we compare these approaches with the EDA Distribution Estimation Using Markov networks (DEUMd). DEUMd constructs a probabilistic model using Markov networks. Our experiments compare the quality of solutions produced by DEUMd with the GA approaches and also reveal interesting differences in the search dynamics that have implications for algorithm design.