ORCID Profile
0000-0003-0397-9923
Current Organisation
Australian Bureau of Statistics
Does something not look right? The information on this page has been harvested from data sources that may not be up to date. We continue to work with information providers to improve coverage and quality. To report an issue, use the Feedback Form.
Publisher: Springer Science and Business Media LLC
Date: 21-03-0008
DOI: 10.1038/S41598-019-41383-Y
Abstract: Predictive risk models using general practice (GP) data to predict the risk of hospitalisation have the potential to identify patients for targeted care. Effective use can help deliver significant reductions in the incidence of hospitalisation, particularly for patients with chronic conditions, the highest consumers of hospital resources. There are currently no published validated risk models for the Australian context using GP data to predict hospitalisation. In addition, published models for other contexts typically rely on a patient’s history of prior hospitalisations, a field not commonly available in GP information systems, as a predictor. We present a predictive risk model developed for use by GPs to assist in targeting coordinated healthcare to patients most in need. The algorithm was developed and validated using a retrospective primary care cohort, linked to records of hospitalisation in Victoria, Australia, to predict the risk of hospitalisation within one year. Predictors employed include demographics, prescription history, pathology results and disease diagnoses. Prior hospitalisation information was not employed as a predictor. Our model shows good performance and has been implemented within primary care practices participating in Health Care Homes, an Australian Government initiative being trialled for providing ongoing comprehensive care for patients with chronic and complex conditions.
Publisher: Elsevier BV
Date: 10-2013
Publisher: Elsevier BV
Date: 06-2005
Publisher: Elsevier BV
Date: 03-2012
DOI: 10.1016/J.JTBI.2011.12.008
Abstract: Hepatitis C virus (HCV) is a blood-borne virus that disproportionately affects people who inject drugs (PWIDs). Based on extensive interview and blood test data from a longitudinal study in Melbourne, Australia, we describe an in idual-based transmission model for HCV spread amongst PWID. We use this model to simulate the transmission of HCV on an empirical social network of PWID. A feature of our model is that sources of infection can be both network neighbours and non-neighbours via "importing". Data-driven estimates of sharing frequency and rate of importing are provided. Compared to an appropriately calibrated fully connected network, the empirical network provides some protective effect on the time to primary infection. We also illustrate heterogeneities in incidence rate of infection, both across and within node degrees (i.e., number of network partners). We explore the reduced risk of infection from spontaneously clearing cutpoint nodes whose infection status oscillates, both in theory and in simulation. Further, we show our model-based estimate of per-event transmission probability largely agrees with previous estimates at the lower end of the range 1-3% commonly cited.
Publisher: Public Library of Science (PLoS)
Date: 11-2013
Publisher: Elsevier BV
Date: 12-2017
Publisher: Wiley
Date: 09-09-2021
Abstract: To determine whether after‐hours presentation to EDs is associated with differences in 7‐day and 30‐day mortality. The influence of patient case‐mix and workforce staffing differences are also explored. We conducted a retrospective observational study of 3.7 million ED episodes across 30 public hospitals in Queensland, Australia during May 2013–September 2015 using routinely collected hospital data linked to hospital staffing data and the death registry. Episodes were categorised as within/after‐hours using time of presentation. Staffing was derived from payroll records and explored by defining 11 staffing ratios. Weekend presentation was slightly more associated (7‐day mortality odds ratio 1.05, 95% confidence interval [CI] 1.01–1.10) or no more associated (30‐day mortality odds ratio 1.01, 95% CI 0.98–1.03) with death than weekday presentation. When weeknights are included in the ‘after‐hours’ period, odds ratios are smaller, so that after‐hours presentation is no more associated (7‐day mortality odds ratio 1.03, 95% CI 0.99–1.08) or less associated (30‐day mortality odds ratio 0.95, 95% CI 0.93–0.97) with death. No significant after‐hours patient case‐mix differences were observed between weekday and weekend presentations for 7‐day mortality. In other combinations of outcome and after‐hours definition, some differences (especially measures relating to severity of presenting condition) were found. Staffing ratios were not strongly associated with any within/after‐hours differences in ED mortality. After‐hours presentation on the weekend to an ED is associated with higher 7‐day mortality even after controlling for case‐mix.
Publisher: Public Library of Science (PLoS)
Date: 10-11-2015
Publisher: Elsevier BV
Date: 02-2019
DOI: 10.1016/J.IJMEDINF.2019.104042
Abstract: To investigate whether the installation of electronic patient journey boards in an inpatient adult rehabilitation centre in Victoria, Australia, is associated with shorter lengths of stay for admitted adult rehabilitation patients. A retrospective before-after analysis of 3 259 adult inpatient rehabilitation episodes from 2013 to 2018 was performed, analysing case-mix adjusted lengths of stay. A reduction in case-mix adjusted length of stay of 4.1 days per episode (95 % confidence interval: 2.0-6.4 days) was found. The corresponding reduction in hospital costs was estimated to be $3 738 per episode (95 % confidence interval $2 398-$4 983). Installation of electronic patient journey boards was associated with shorter lengths of stay in an inpatient adult rehabilitation centre. Additional research is needed to 1) provide further evidence of the causal effect of the boards on length of stay, and 2) investigate the mechanisms by which they reduce lengths of stay (e.g., increased currency of information, changes to procedures, remote viewing) in rehabilitation settings.
Publisher: Institute of Mathematical Statistics
Date: 2011
DOI: 10.1214/ECP.V16-1673
Publisher: Springer Science and Business Media LLC
Date: 02-11-2015
Publisher: Elsevier BV
Date: 10-2016
Publisher: Elsevier BV
Date: 10-2015
DOI: 10.1016/J.DRUGPO.2015.05.006
Abstract: The hepatitis C virus (HCV) epidemic is a major health issue in most developed countries it is driven by people who inject drugs (PWID). Injecting networks powerfully influence HCV transmission. In this paper we provide an overview of 10 years of research into injecting networks and HCV, culminating in a network-based approach to provision of direct-acting antiviral therapy. Between 2005 and 2010 we followed a cohort of 413 PWID, measuring HCV incidence, prevalence and injecting risk, including network-related factors. We developed an in idual-based HCV transmission model, using it to simulate the spread of HCV through the empirical social network of PWID. In addition, we created an empirically grounded network model of injecting relationships using exponential random graph models (ERGMs), allowing simulation of realistic networks for investigating HCV treatment and intervention strategies. Our empirical work and modelling underpins the TAP Study, which is examining the feasibility of community-based treatment of PWID with DAAs. We observed incidence rates of HCV primary infection and reinfection of 12.8 per 100 person-years (PY) (95%CI: 7.7-20.0) and 28.8 per 100 PY (95%CI: 15.0-55.4), respectively, and determined that HCV transmission clusters correlated with reported injecting relationships. Transmission modelling showed that the empirical network provided some protective effect, slowing HCV transmission compared to a fully connected, homogenous PWID population. Our ERGMs revealed that treating PWID and all their contacts was the most effective strategy and targeting treatment to infected PWID with the most contacts the least effective. Networks-based approaches greatly increase understanding of HCV transmission and will inform the implementation of treatment as prevention using DAAs.
Publisher: Ovid Technologies (Wolters Kluwer Health)
Date: 24-10-2014
DOI: 10.1002/HEP.27403
Abstract: With the development of new highly efficacious direct-acting antiviral (DAA) treatments for hepatitis C virus (HCV), the concept of treatment as prevention is gaining credence. To date, the majority of mathematical models assume perfect mixing, with injectors having equal contact with all other injectors. This article explores how using a networks-based approach to treat people who inject drugs (PWID) with DAAs affects HCV prevalence. Using observational data, we parameterized an exponential random graph model containing 524 nodes. We simulated transmission of HCV through this network using a discrete time, stochastic transmission model. The effect of five treatment strategies on the prevalence of HCV was investigated two of these strategies were (1) treat randomly selected nodes and (2) "treat your friends," where an in idual is chosen at random for treatment and all their infected neighbors are treated. As treatment coverage increases, HCV prevalence at 10 years reduces for both the high- and low-efficacy treatment. Within each set of parameters, the treat your friends strategy performed better than the random strategy being most marked for higher-efficacy treatment. For ex le, over 10 years of treating 25 per 1,000 PWID, the prevalence drops from 50% to 40% for the random strategy and to 33% for the treat your friends strategy (6.5% difference 95% confidence interval: 5.1-8.1). Treat your friends is a feasible means of utilizing network strategies to improve treatment efficiency. In an era of highly efficacious and highly tolerable treatment, such an approach will benefit not just the in idual, but also the community more broadly by reducing the prevalence of HCV among PWID.
Publisher: Institute of Mathematical Statistics
Date: 11-2010
DOI: 10.1214/09-BJPS105
Publisher: Public Library of Science (PLoS)
Date: 26-10-2012
Publisher: Springer Science and Business Media LLC
Date: 10-12-2021
DOI: 10.1038/S41598-021-02827-6
Abstract: To improve understanding of Alzheimer’s disease, large observational studies are needed to increase power for more nuanced analyses. Combining data across existing observational studies represents one solution. However, the disparity of such datasets makes this a non-trivial task. Here, a machine learning approach was applied to impute longitudinal neuropsychological test scores across two observational studies, namely the Australian Imaging, Biomarkers and Lifestyle Study (AIBL) and the Alzheimer's Disease Neuroimaging Initiative (ADNI) providing an overall harmonised dataset. MissForest, a machine learning algorithm, capitalises on the underlying structure and relationships of data to impute test scores not measured in one study aligning it to the other study. Results demonstrated that simulated missing values from one dataset could be accurately imputed, and that imputation of actual missing data in one dataset showed comparable discrimination (p 0.001) for clinical classification to measured data in the other dataset. Further, the increased power of the overall harmonised dataset was demonstrated by observing a significant association between CVLT-II test scores (imputed for ADNI) with PET Amyloid-β in MCI APOE -ε4 homozygotes in the imputed data (N = 65) but not for the original AIBL dataset (N = 11). These results suggest that MissForest can provide a practical solution for data harmonization using imputation across studies to improve power for more nuanced analyses.
Publisher: Springer Science and Business Media LLC
Date: 05-10-2022
DOI: 10.1038/S41598-022-20907-Z
Abstract: Preventing unplanned hospitalisations, including readmissions and re-presentations to the emergency department, is an important strategy for addressing the growing demand for hospital care. Significant successes have been reported from interventions put in place by hospitals to reduce their incidence. However, there is limited use of data-driven algorithms in hospital services to identify patients for enrolment into these intervention programs. Here we present the results of a study aiming to develop algorithms deployable at scale as part of a state government’s initiative to address rehospitalizations and which fills several gaps identified in the state-of-the-art literature. To the best of our knowledge, our study involves the largest-ever s le size for developing risk models. Logistic regression, random forests and gradient boosted techniques were explored as model candidates and validated retrospectively on five years of data from 27 hospitals in Queensland, Australia. The models used a range of predictor variables sourced from state-wide Emergency Department(ED), inpatient, hospital-dispensed medications and hospital-requested pathology databases. The investigation leads to several findings: (i) the advantage of looking at a longer patient data history, (ii) ED and inpatient datasets alone can provide useful information for predicting hospitalisation risk and the addition of medications and pathology test results leads to trivial performance improvements, (iii) predicting readmissions to the hospital was slightly easier than predicting re-presentations to ED after an inpatient stay, which was slightly easier again than predicting re-presentations to ED after an EDstay, ( iv ) a gradient boosted approach (XGBoost) was systematically the most powerful modelling approach across various tests.
Publisher: Wiley
Date: 03-2011
Publisher: Informa UK Limited
Date: 17-11-2015
Location: United States of America
Location: Australia
No related grants have been discovered for David Rolls.