Queensland Government

Objectives To develop and validate an algorithm for estimating EQ-5D-5L utility scores from the nonpreference-based Healthy Days (HD) instrument using direct and indirect mapping approaches. Methods Data were drawn from the Queensland Preventive Health Survey for 2022 (N = 12 491) and 2023 (N = 12 679). Survey-weighted ordinary least squares (OLS), Tobit, beta, majorize-minimize, and ordinal logistic regression models were evaluated. Algorithms were developed using an 80% random training sample from the 2022 data, tested in the remaining 20%, and externally validated using 2023 data. Predictive performance was assessed using root mean square error, mean absolute error, differences between predicted and observed mean utilities, and subgroup analyses by age, sex, body mass index, smoking status, and utility quartile. Results Moderate correlations were observed between unhealthy days measures and EQ-5D-5L utilities (Pearson’s r: −0.53 to −0.56). Predictive accuracy was broadly similar for the OLS, beta, and ordinal logistic models. The OLS model including physically unhealthy days, mentally unhealthy days, activity limitation days, self-rated health, age, and sex was selected as the preferred algorithm. It demonstrated strong and stable predictive performance across samples (root mean square error ≤ 0.117 and mean absolute error ≤ 0.066 in validation samples). All models overestimated utilities in the lowest utility quartile. Conclusions The HD instrument can be mapped to the EQ-5D-5L with favorable predictive accuracy in population survey data. The proposed algorithm provides a pragmatic approach for estimating mean utilities for quality-adjusted life-year calculation when only HD data are available, although caution is warranted in populations with very low baseline health status.

This study aimed to evaluate how shifts in land use and climate extremes have jointly shaped long-term water quality trends in the Sacramento River watershed in California between 1985 and 2023. We integrated satellite-derived land use maps, daily climate indices, and monthly water quality records for six parameters (hardness, pH, specific conductance, total dissolved solids (TDS), total suspended solids (TSS), and turbidity). Analyses included seasonal trend assessments, spectrogram analysis of periodicity, and Random Forest models to identify key drivers of water quality trends. Three modeling cases were compared: climate-only, land use-only, and combined (climate + land use). A 10-fold cross-validation framework was implemented, and differences among models were assessed using the Friedman test with Wilcoxon signed-rank post-hoc comparisons. The combined model integrating climate and land use variables best explained long-term water quality variability, followed by land use alone and climate alone. Seasonal analysis further highlighted a dual stress regime: winter storms drove sediment and turbidity peaks, while summer droughts intensified salinity and pH. Water quality deterioration peaked during 2002-2012, with elevated salinity (conductance, TDS) and sediment-related parameters (TSS, turbidity). Land use predictors explain more site-to-site variation in water quality than climate predictors (median absolute ΔR2 LU - Clim ≈ 0.18; LU outperforms Clim for 5/6 variables). Persistence of elevated salinity underscores the need for long-term management, while the partial recovery of sediment indicators highlights opportunities for targeted erosion control. These findings show the interacting influence of land use transitions and climate extremes, offering a transferable framework for adaptive watershed management.

Hyperspectral images (HSIs) have demonstrated remarkable potential in remote sensing applications due to their rich spectral and spatial characteristics. However, the scarcity of labeled HSI data from individual sources significantly hinders the deployment of deep learning models in this domain. Leveraging multisource HSI can mitigate single-source data limitations, but direct data sharing poses serious privacy challenges. Federated learning (FL) provides a compelling framework for privacy-preserving collaborative learning, yet practical challenges remain, particularly data scarcity and the nonindependent and identically distributed (non-IID) nature of client data. In this work, we introduce a diffusion-based data augmentation framework, namely FedDA-HSI, tailored for federated HSI (FedHSI) classification. FedDA-HSI addresses two core issues in real-world FedHSI deployments. First, we propose a federated diffusion model capable of generating high-fidelity synthetic hyperspectral data locally, enabling data augmentation without compromising privacy, while allowing the aggregated global model to capture shared knowledge across clients. Second, we incorporate a class-aware augmentation strategy to alleviate data imbalance by injecting synthetic samples of missing or underrepresented classes, effectively improving interclient data diversity and mitigating non-IID effects. Extensive experiments on benchmark HSI datasets validate that our approach significantly improves classification performance under heterogeneous conditions.

People with chronic diseases are known to have lower EQ-5D-5L utility scores, but data are not readily available in an Australian context. Using linked administrative hospital and population survey data, we aimed to calculate utility scores for adults with different disease profiles. We conducted a retrospective cohort study using a cross-sectional population-level health survey (2022–2023) linked to administrative hospital data for adults 18 years and older in Queensland, Australia to assess: (1) chronic disease and comorbidity prevalence, (2) Health-related quality of life (HRQoL) differences among adults with pre-existing chronic conditions, and (3) to model differences in disutility days by chronic diseases. The mean EQ-5D-5L utility score for the cohort was 0.917, but was lower among those with chronic diseases, for example chronic obstructive pulmonary disease (COPD), coronary heart disease (CHD) and diabetes had corresponding mean values of 0.780, 0.850 and 0.832, respectively. After adjustment, on average the disutility days that could be averted by preventing chronic diseases equalled approximately a month annually for some conditions, ranging from 14.8 days for CHD to 36.5 days for COPD. Prevalence estimates using linked administrative hospital data were comparable to results from the National Health Survey, which used self-report, although comorbidity was found to be substantially higher in the current study. People living with chronic diseases have substantially higher number of disutility days annually. Preventing or delaying onset of chronic conditions would likely improve HRQoL and positively impact individuals, society and the economy. Why is this study needed? Health-related quality of life (HRQoL) is lower among those with chronic diseases than those without. While there are methods to evaluate HRQoL numerically, estimates for people living with chronic diseases are not available in an Australian context. What is the key problem/issue/question this manuscript addresses? The key questions this manuscript addresses are to: (1) assess chronic disease and comorbidity prevalence using population-level survey data linked to administrative health records, (2) quantify HRQoL differences among adults with pre-existing chronic conditions, and (3) model differences in disutility days by listed chronic diseases. What is the main point of your study? Using a population health survey linked to hospital admission and emergency presentation data, this study estimated the prevalence of selected chronic diseases in Queensland, Australia, and evaluated differences in HRQoL scores and the number of days not in optimal health (disutility day) among people with different disease profiles. What are your main results and what do they mean? Survey data linked to administrative hospital data provided chronic disease prevalence estimates comparable to a national health survey, although multimorbidity estimates were higher. Multimorbidity is commonly associated with higher health care costs and utilisation, and is a growing challenge to healthcare sector sustainability. Robust multimorbidity prevalence better informs healthcare system planning. Population-level HRQoL scores were also lower among Queensland adults with chronic diseases. A high number of disutility days could be avoided if chronic disease onset could be averted or delayed.

Gross Primary Production (GPP) is a fundamental component of the ecosystem carbon cycle, and its accurate estimation is critical for understanding the global carbon budget and the ecosystem’s response to climate change. Theoretically, increasing the number of observation sites and extending the observational period should improve the predictive accuracy of machine learning models. However, in practice, model performance improvements are not strictly linear and may exhibit saturation, where additional data provide diminishing returns. Based on GPP flux observations across Europe from 2001 to 2020, this study conducted two comparative experiments to systematically assess the impact of data expansion on GPP estimation accuracy. A plant functional types (PFTs) classification strategy was further introduced to explore its potential for improving model performance. The results show that extending the temporal range from 2001 to 2020 only slightly increased R ^2 from 0.722 to 0.725. Spatial expansion further reduced accuracy, with R ^2 dropping from 0.740 to 0.734. In contrast, incorporating PFT-based modeling significantly enhanced estimation performance, raising R ^2 to approximately 0.77 (p < 0.01) and reducing prediction errors. However, changes in data volume have an impact on the spatiotemporal pattern of GPP estimates by the model, with significant inconsistencies in trends and seasonal dynamics across regions and vegetation types. These findings highlight the limitations of current machine learning models and the fact that, as data availability increases, improving the accuracy of GPP estimates will rely more on optimized model structure and ecological stratification rather than simply on data volume.