Flinders Digital Health Research Centre Collected Works

Permanent URI for this collection

https://hdl.handle.net/2328/36741

Browse

Now showing 1 - 20 of 39

On the arguments against the application of data mining to medical data analysis
(IDAMAP, 2006) Shillabeer, Annette; Roddick, John Francis; de Vries, Denise Bernadette
There is a variety of criticisms of medical data mining which has led, in some cases, to the technology being overlooked as a tool. This paper presents a discussion of six of the strongest arguments against the application of data mining to the complex field of human medicine. The aim of the paper is to raise the predominant issues and suggest solutions whilst also opening the issues for further consideration by both medical and information technology communities.
Simulated Artificial Human Vision: The Effects of Spatial Resolution and Frame Rate on Mobility
(IOS Press, 2006) Dowling, Jason A; Boles, Wageeh; Maeder, Anthony
Electrical stimulation of the human visual system can result in the perception of blobs of light, known as phosphenes. Artificial Human Vision (AHV or visual prosthesis) systems use this method to provide a visual substitute for the blind. This paper reports on our experiments involving normally sighted participants using a portable AHV simulation. A Virtual Reality Head Mounted Display is used to display the phosphene simulation. Custom software converts captured images from a head mounted USB camera to a DirectX based phosphene simulation. The effects of frame rate (1, 2 and 4 FPS) and phosphene spatial resolution (16x12 and 32x24) on participant Percentage of Preferred Walking Speed (PPWS) and mobility errors were assessed during repeated trials on an artificial indoor mobility course. Results indicate that spatial resolution is a significant factor in reducing contact with obstacles and following a path without veering, however the phosphene display frame rate is a better predictor of a person’s preferred walking speed. These findings support the development of an adaptive display which could provide a faster display with reduced spatial resolution when a person is walking comfortably and a slower display with higher resolution when a person has stopped moving.
Towards Role Based Hypothesis Evaluation for Health Data Mining
(2006) Shillabeer, Annette; Roddick, John Francis
Data mining researchers have long been concerned with the application of tools to facilitate and improve data analysis on large, complex data sets. The current challenge is to make data mining and knowledge discovery systems applicable to a wider range of domains, among them health. Early work was performed over transactional, retail based data sets, but the attraction of finding previously unknown knowledge from the ever increasing amounts of data collected from the health domain is an emerging area of interest and specialisation. The problem is finding a solution that is suitably flexible to allow for generalised application whilst being specific enough to provide functionality that caters for the nuances of each role within the domain. The need for a more granular approach to problem solving in other areas of information technology has resulted in the use of role based solutions. This paper discusses the progress to date in developing a role oriented solution to the problem of providing for the diverse requirements of health domain data miners and defining the foundation for determining what constitutes an interesting discovery in an area as complex as health.
Establishing a lineage for medical knowledge discovery
(Australian Computer Society, 2007) Roddick, John Francis; Shillabeer, Annette
Medical science has a long history characterised by incidents of extraordinary insights that have resulted in a paradigm shift in the methodologies and approaches used and have moved the discipline forward. While knowledge discovery has much to offer medicine, it cannot be done in ignorance of either this history or the norms of modern medical investigation. This paper explores the lineage of medical knowledge acquisition and discusses the adverse perceptions that data mining techniques will have to surmount to gain acceptance.
e-Research Meets e-Health
(Australian Computer Society, 2008) Maeder, Anthony
This paper considers some key aspects of e-Research methodology and infrastructure which are relevant to e-Health, and identifies some promising areas in e-Health where these aspects could be used beneficially .
Discovering itemset interactions
(Australian Computer Society, 2009) Liang, Ping; Roddick, John Francis; Ceglar, Aaron John; Shillabeer, Anna; de Vries, Denise Bernadette
Itemsets, which are treated as intermediate results in association mining, have attracted significant research due to the inherent complexity of their generation. However, there is currently little literature focusing upon the interactions between itemsets, the nature of which may potentially contain valuable information. This paper presents a novel tree-based approach to discovering item-set interactions, a task which cannot be undertaken by current association mining techniques.
Assessing Viewing Pattern Consistency in Mammogram Readers
(Australian Computer Society, 2009) Maeder, Anthony; Fookes, C
Breast cancer screening programs typically require very large volumes of x-ray images (mammograms) to be viewed by highly experienced human readers. The readers can recognise a wide range of different visible features indicative of clinically abnormal situations, which they use as a basis to generate a report on their findings. Errors in reporting can occur if the readers fail to identify a particular feature of interest for further visual inspection during the viewing process. This risk is typically reduced by training readers to follow a particular viewing path through an image, which they should be able to apply consistently. Knowledge of the extent of consistency in this viewing behaviour within and between viewers would inform the development of an automated checking approach, based on monitoring of viewer visual attention. This paper presents an analysis of some reader viewing pattern profiles obtained using eye tracking with an infra red computer vision system, as a basis for developing a suitable consistency assessment model. .
Characterizing Image Properties for Digital Mammograms
(Australian Computer Society, 2009) Nguyen, Anthony; Dowling, Jason A; Maeder, Anthony; Nguyen, Phuong; Brunton, Emma
Adoption of computed radiology (CR) and direct radiology (DR) imaging for screening mammograms in many countries alongside digitally scanned film mammograms has resulted in a wide range of different intrinsic (physical) characteristics of images becoming commonplace. It is sometimes conjectured that viewer performance could be adversely affected by this wider variability, as compared with the variability that was formerly experienced with film only. This paper identifies several aspects of the image characteristics relevant to viewer perception, including intensity properties (such as contrast), spatial properties (such as texture) and structure properties (such as breast density). We then provide quantitative descriptions of the variability of these properties over a test set of 12 screening mammograms drawn from three different modalities and containing a typical mix of screening cases..
Gaining insight from patient journey data using process-oriented analysis approach
(Australian Computer Society, 2012) Perimal-Lewis, Lua; Qin, Shaowen; Thompson, Campbell Henry; Hakendorf, Paul Haylett
Hospitals are continually struggling to cater for the increasing demand for inpatient services. This is due to increased population, aging, and the rising incidence of chronic diseases associated with modern life. The high demand for hospital services leads to unpredictable bed availability, longer waiting period for acute admission, difficulties in keeping planned admission, stressed hospital staff, undesirable patient and family experience, as well as unclear long term impact on health care capacity. This study aims to derive some correlation between various factors contributing to ward occupancy rate and operation efficiency. The aim is also to discover the inpatient flow process model proposing to use process mining techniques combined with data analysis to depict the relationships among inpatients, wards and Length of Stay (LOS) in an effort to gain insight into factors that could be focused to relieve access block. Open source process mining software - ProM is used for this study. The study is done in collaboration with Flinders Medical Centre (FMC) using data from their Patient Journey Database as case study.
Constructing a Synthetic Longitudinal Health Dataset for Data Mining
(IARIA, 2012) Ghassem Pour, S; Maeder, Anthony; Jorm, L
The traditional approach to epidemiological research is to analyse data in an explicit statistical fashion, attempting to answer a question or test a hypothesis. However, increasing experience in the application of data mining and exploratory data analysis methods suggests that valuable information can be obtained from large datasets using these less constrained approaches. Available data mining techniques, such as clustering, have mainly been applied to cross-sectional point-in-time data. However, health datasets often include repeated observations for individuals and so researchers are interested in following their health trajectories. This requires methods for analysis of multiple-points-over-time or longitudinal data. Here, we describe an approach to construct a synthetic longitudinal version of a major population health dataset in which clusters merge and split over time, to investigate the utility of clustering for discovering time sequence based patterns.
Comparing Data Mining with Ensemble Classiﬁcation of Breast Cancer Masses in Digital Mammograms
(AIH, 2012) Ghassem Pour, S; McLeod, P; Verma, B; Maeder, Anthony
Medical diagnosis sometimes involves detecting subtle indi-cations of a disease or condition amongst a background of diverse healthy individuals. The amount of information that is available for discover-ing such indications for mammography is large and has been growing at an exponential rate, due to population wide screening programmes. In order to analyse this information data mining techniques have been utilised by various researchers. A question that arises is: do ﬂexible data mining techniques have comparable accuracy to dedicated classiﬁcation techniques for medical diagnostic processes? This research compares a model-based data mining technique with a neural network classiﬁcation technique and the improvements possible using an ensemble approach. A publicly available breast cancer benchmark database is used to determine the utility of the techniques and compare the accuracies obtained.
Effectiveness of a website and mobile phone based physical activity and nutrition intervention for middle-aged males: Trial protocol and baseline findings of the ManUp Study
(BioMed Central, 2012-08-15) Duncan, Mitch J; Vandelanotte, Corneel; Rosenkranz, Richard R; Caperchione, Cristina M; Ding, Hang; Ellison, Marcus; George, Emma S; Hooker, Cindy; Karunanithi, Mohan; Kolt, Gregory S; Maeder, Anthony; Noakes, Manny; Tague, Rhys; Taylor, Pennie; Viljoen, Pierre; Mummery, W Kerry
Background: Compared to females, males experience higher rates of chronic disease and mortality, yet few health promotion initiatives are specifically aimed at men. Therefore, the aim of the ManUp Study is to examine the effectiveness of an IT-based intervention to increase the physical activity and nutrition behaviour and literacy in middle-aged males (aged 35–54 years). Method/Design: The study design was a two-arm randomised controlled trial, having an IT-based (applying website and mobile phones) and a print-based intervention arm, to deliver intervention materials and to promote self-monitoring of physical activity and nutrition behaviours. Participants (n = 317) were randomised on a 2:1 ratio in favour of the IT-based intervention arm. Both intervention arms completed assessments at baseline, 3, and 9 months. All participants completed self-report assessments of physical activity, sitting time, nutrition behaviours, physical activity and nutrition literacy, perceived health status and socio-demographic characteristics. A randomly selected sub-sample in the IT-based (n = 61) and print-based (n = 30) intervention arms completed objective measures of height, weight, waist circumference, and physical activity as measured by accelerometer (Actigraph GT3X). The average age of participants in the IT-based and print-based intervention arm was 44.2 and 43.8 years respectively. The majority of participants were employed in professional occupations (IT-based 57.6%, Print-based 54.2%) and were overweight or obese (IT-based 90.8%, Print-based 87.3%). At baseline a lower proportion of participants in the IT-based (70.2%) group agreed that 30 minutes of physical activity each day is enough to improve health compared to the print-based (82.3%) group (p = .026). The IT-based group consumed a significantly lower number of serves of red meat in the previous week, compared to the print-based group (p = .017). No other significant between-group differences were observed at baseline. Discussion: The ManUp Study will examine the effectiveness of an IT-based approach to improve physical activity and nutrition behaviour and literacy. Study outcomes will provide much needed information on the efficacy of this approach in middle aged males, which is important due to the large proportions of males at risk, and the potential reach of IT-based interventions.
The relationship between in-hospital location and outcomes of care in patients of a large general medical service
(Wiley-Blackwell, 2012-12-24) Perimal-Lewis, Lua; Li, Jordan Yuanzhi; Hakendorf, Paul Haylett; Ben-Tovim, David Isaac; Qin, Shaowen; Thompson, Campbell Henry
Background: The discrepancy between the number of admissions and the allocation of hospital beds means many patients admitted under the care of a general medical service can be placed in other departments’ wards. These patients are called “outliers” and their outcomes are unknown. Aims: To examine the relation between the proportion of time each patient spent in their “home ward” during an index admission and the outcomes of that hospital stay. Methods: Data from Flinders Medical Centre’s (FMC) patient journey database were extracted and analysed. The analysis was carried out on the patient journeys of patients admitted under the General Medicine units. Results: Outlier patients’ length of stay (LOS) was significantly shorter than that of the inlier patients (110.7 hours cf 141.9 hours; p < 0.001).They had a reduced risk of readmission within 28 days of discharge from hospital. Outlier patients’ discharge summaries were less likely to be completed within a week (64.3% cf 78.0%; p < 0.001). Being an outlier patient increased the risk-adjusted risk of in-hospital mortality by over 40%. 50% of deaths in the outlier group occurred within 48 hours of admission. Outlier patients had spent longer in the Emergency Department (ED) waiting for a bed (6.3 hours cf 5.3 hours; p < 0.001) but duration of ED stay was not an independent predictor of mortality risk. Conclusion: Outlier patients had significantly shorter LOS in hospital, but significantly greater in-patient death rates. Surviving outlier patients had lower rates of readmission but lower rates of discharge summary completion.
A conceptual framework for secure mobile health
(International Society for Telemedicine and eHealth, 2013) William, P; Maeder, Anthony
Mobile health is characterised by its diversity of applicability, in a multifaceted and multidisciplinary healthcare delivery continuum. In an environment of rapid change with the increasing development of mobile health, issues related to security and privacy must be well thought out. The different competing tensions in the development of mobile health from the device technologies and associated regulation, to clinical workflow and patient acceptance, require a framework for security that reflects the complex structure of this emerging field. There are three distinct associated elements that require investigation: technology, clinical, and human factors. Each of these elements consists of multiple aspects and there are specific risk factors to be addressed successively and co-dependently in each case. The fundamental approach to defining a conceptual framework for secure use of mobile health requires systematic identification of properties for the tensions and critical factors which impact these elements. The resulting conceptual framework presented here can be used for new critique, augmentation or deployment of mobile health solutions from the perspective of data protection and security.
Validating Synthetic Health Datasets for Longitudinal Clustering
(2013) Ghassem Pour, S; Maeder, Anthony; Jorm, L
Clustering methods partition datasets into subgroups with some homogeneous properties, with information about the number and particular characteristics of each subgroup unknown a priori. The problem of predicting the number of clusters and quality of each cluster might be overcome by using cluster validation methods. This paper presents such an approach in-corporating quantitative methods for comparison be-tween original and synthetic versions of longitudinal health datasets. The use of the methods is demon-strated by using two diﬀerent clustering algorithms, K-means and Latent Class Analysis, to perform clus-tering on synthetic data derived from the 45 and Up Study baseline data, from NSW in Australia.
Analysing homogenous patient journeys to assess quality of care for patients admitted outside of their ‘home-ward’
(Australian Computer Society, 2013) Perimal-Lewis, Lua; Qin, Shaowen; Thompson, Campbell Henry; Hakendorf, Paul Haylett
This study is the first to explore the quality of care based on the outlier or the inlier status of patients for a large heterogeneous General Medicine (GM) service at a busy public hospital. The study compared the quality of care between ward outliers and ward inliers based on a homogenous group of patients using Two-step clustering method. Contrary to common perception, ward outliers had overall shorter Length of Stay (LOS) than ward inliers. The study also was unable to support the perception of shorter LOS in the outlier group being associated with higher in-hospital mortality. The study confirmed that overall the outliers received inferior quality of care as discharge summaries for the outliers were delayed and more outliers were re-admitted within 7 days of discharge in comparison to the inliers.
WALK 2.0 - Using Web 2.0 applications to promote health-related physical activity: A randomised controlled trial protocol
(BioMed Central, 2013-05-03) Kolt, Gregory S; Rosenkranz, Richard R; Savage, Trevor N; Maeder, Anthony; Vandelanotte, Corneel; Duncan, Mitch J; Caperchione, Cristina M; Tague, Rhys; Hooker, Cindy; Mummery, W Kerry
Background: Physical inactivity is one of the leading modifiable causes of death and disease in Australia. National surveys indicate less than half of the Australian adult population are sufficiently active to obtain health benefits. The Internet is a potentially important medium for successfully communicating health messages to the general population and enabling individual behaviour change. Internet-based interventions have proven efficacy; however, intervention studies describing website usage objectively have reported a strong decline in usage, and high attrition rate, over the course of the interventions. Web 2.0 applications give users control over web content generated and present innovative possibilities to improve user engagement. There is, however, a need to assess the effectiveness of these applications in the general population. The Walk 2.0 project is a 3-arm randomised controlled trial investigating the effects of “next generation” web-based applications on engagement, retention, and subsequent physical activity behaviour change. Methods/design: 504 individuals will be recruited from two sites in Australia, randomly allocated to one of two web-based interventions (Web 1.0 or Web 2.0) or a control group, and provided with a pedometer to monitor physical activity. The Web 1.0 intervention will provide participants with access to an existing physical activity website with limited interactivity. The Web 2.0 intervention will provide access to a website featuring Web 2.0 content, including social networking, blogs, and virtual walking groups. Control participants will receive a logbook to record their steps. All groups will receive similar educational material on setting goals and increasing physical activity. The primary outcomes are objectively measured physical activity and website engagement and retention. Other outcomes measured include quality of life, psychosocial correlates, and anthropometric measurements. Outcomes will be measured at baseline, 3, 12 and 18 months. Discussion: The findings of this study will provide increased understanding of the benefit of new web-based technologies and applications in engaging and retaining participants on web-based intervention sites, with the aim of improved health behaviour change outcomes.
Effectiveness of a Web- and Mobile Phone-Based Intervention to Promote Physical Activity and Healthy Eating in Middle-Aged Males: Randomized Controlled Trial of the ManUp Study
(Journal of Medical Internet Research, 2014) Duncan, Mitch J; Vandelanotte, Corneel; Kolt, Gregory S; Rosenkranz, Richard R; Caperchione, Cristina M; George, Emma S; Ding, Hang; Hooker, Cindy; Karunanithi, Mohan; Maeder, Anthony; Noakes, Manny; Tague, Rhys; Taylor, Pennie; Viljoen, Pierre; Mummery, W Kerry
Background: The high number of adult males engaging in low levels of physical activity and poor dietary practices, and the health risks posed by these behaviors, necessitate broad-reaching intervention strategies. Information technology (IT)-based (Web and mobile phone) interventions can be accessed by large numbers of people, yet there are few reported IT-based interventions targeting males’ physical activity and dietary practices. Objective: This study examines the effectiveness of a 9-month IT-based intervention (ManUp) to improve the physical activity, dietary behaviors, and health literacy in middle-aged males compared to a print-based intervention. Methods: Participants, recruited offline (eg, newspaper ads), were randomized into either an IT-based or print-based intervention arm on a 2:1 basis in favor of the fully automated IT-based arm. Participants were adult males aged 35-54 years living in 2 regional cities in Queensland, Australia, who could access the Internet, owned a mobile phone, and were able to increase their activity level. The intervention, ManUp, was based on social cognitive and self-regulation theories and specifically designed to target males. Educational materials were provided and self-monitoring of physical activity and nutrition behaviors was promoted. Intervention content was the same in both intervention arms; only the delivery mode differed. Content could be accessed throughout the 9-month study period. Participants’ physical activity, dietary behaviors, and health literacy were measured using online surveys at baseline, 3 months, and 9 months. Results: A total of 301 participants completed baseline assessments, 205 in the IT-based arm and 96 in the print-based arm. A total of 124 participants completed all 3 assessments. There were no significant between-group differences in physical activity and dietary behaviors (P≥.05). Participants reported an increased number of minutes and sessions of physical activity at 3 months (exp(β)=1.45, 95% CI 1.09-1.95; exp(β)=1.61, 95% CI 1.17-2.22) and 9 months (exp(β)=1.55, 95% CI 1.14-2.10; exp(β)=1.51, 95% CI 1.15-2.00). Overall dietary behaviors improved at 3 months (exp(β)=1.07, 95% CI 1.03-1.11) and 9 months (exp(β)=1.10, 95% CI 1.05-1.13). The proportion of participants in both groups eating higher-fiber bread and low-fat milk increased at 3 months (exp(β)=2.25, 95% CI 1.29-3.92; exp(β)=1.65, 95% CI 1.07-2.55). Participants in the IT-based arm were less likely to report that 30 minutes of physical activity per day improves health (exp(β)=0.48, 95% CI 0.26-0.90) and more likely to report that vigorous intensity physical activity 3 times per week is essential (exp(β)=1.70, 95% CI 1.02-2.82). The average number of log-ins to the IT platform at 3 and 9 months was 6.99 (SE 0.86) and 9.22 (SE 1.47), respectively. The average number of self-monitoring entries at 3 and 9 months was 16.69 (SE 2.38) and 22.51 (SE 3.79), respectively. Conclusions: The ManUp intervention was effective in improving physical activity and dietary behaviors in middle-aged males with no significant differences between IT- and print-based delivery modes.
Health intelligence: Discovering the process model using process mining by constructing Start-to-End patient journeys
(Australian Computer Society, 2014) Perimal-Lewis, Lua; de Vries, Denise Bernadette; Thompson, Campbell Henry
Australian Public Hospitals are continually engaged in various process improvement activities to improve patient care and to improve hospital efficiency as the demand for service intensifies. As a consequence there are many initiatives within the health sector focusing on gaining insight into the underlying health processes which are assessed for compliance with specified Key Performance Indicators (KPIs). Process Mining is classified as a Business Intelligence (BI) tool. The aim of process mining activities is to gain insight into the underlying process or processes. The fundamental element needed for process mining is a historical event log of a process. Generally, these event logs are easily sourced from Process Aware Information Systems (PAIS). Simulation is widely used by hospitals as a tool to study the complex hospital setting and for prediction. Generally, simulation models are constructed by ‘hand’. This paper presents a novel way of deriving event logs for health data in the absence of PAIS. The constructed event log is then used as an input for process mining activities taking advantage of existing process mining algorithms aiding the discovery of knowledge of the underlying processes which leads to Health Intelligence (HI). One such output of process mining activity, presented in this paper, is the discovery of process model for simulation using the derived event log as an input for process mining by constructing start-to-end patient journey. The study was undertaken using data from Flinders Medical Centre to gain insight into patient journeys from the point of admission to the Emergency Department (ED) until the patient is discharged from the hospital. .
Evaluating success of mobile health projects in the developing world
(IOS Press, 2014) Ginige, J A; Maeder, Anthony; Long, V
Many mobile health (mHealth) projects, typically deploying pilot or small scale implementations, have been undertaken in developing world settings and reported with a widely varying range of claims being made on their effectiveness and benefits. As a result, there is little evidence for which aspects of such projects lead to successful outcomes. This paper describes a literature review of papers from PubMed undertaken to identify strong contributions to execution and evaluation of mHealth projects in developing world settings, and suggests a template for classifying the main success factors to assist with collating evidence in the future.

Browse

Browsing Flinders Digital Health Research Centre Collected Works by Issue Date

Results Per Page

Sort Options