Studies in health technology and informatics最新文献

Eating disorders are complex mental health conditions with significant morbidity and mortality, and a rising prevalence in children and adolescents. Despite global research and clinical efforts to improve outcomes, establishing routine, longitudinal data collection that facilitates individualised care in real-time and outcome assessment across clinical cohorts is crucial. The Royal Children's Hospital Eating Disorders Service have developed a world-first evaluation program that is integrated into the electronic medical record in a paediatric setting, capturing demographics, clinical characteristics, and treatment outcomes of young people receiving care over time. The design can be scaled across services to expand the dataset and enable comparisons of treatment modalities and subgroup outcomes - improving clinical decision making and, enabling longitudinal data collection, and facilitating national and international collaboration.

This study explores Australian consumers' digital literacy (DL), use of digital health technologies (DHTs), and curiosity toward emerging tools. A cross-sectional online survey (n = 416) examined DL levels, current usage of technologies such as telehealth, wearables, mHealth apps, e-pharmacy, and chatbots, and preferences for future innovations like smart glasses, virtual reality/augmented reality, medical drones, and robot companions. DL was highest in data and communication domains and varied by age, gender, education, and location. Despite women and younger adults reporting higher DL, technology adoption often hinged on perceived usefulness, usability, and trust. Telehealth was widely used (90%+) while emerging technologies attracted greater curiosity from men and the 30-39 age group. These findings suggest that curiosity - both diversive and specific - drives early exploration and continued engagement with DHTs. To support equitable adoption, digital health strategies should integrate DL-building interventions and curiosity-driven design, aligned with the Australian Digital Health Strategy's goals for inclusive, consumer-centred innovation.

Accurate hospital bed demand forecasting is critical for ensuring effective patient care and efficient resource allocation. This study evaluates various statistical and machine learning methods to predict daily and hourly inpatient admissions, separations, and emergency department (ED) presentations up to one year in advance. The Advanced Demand Prediction Tool (ADePT) is introduced, which leverages the SARIMAX time series model to capture trends, seasonal patterns, and public holiday effects. Its performance is evaluated using data from a large provider of tertiary health services in Melbourne, Australia against five other statistical and machine learning forecasting models, including rolling window, six-week rolling average, negative binomial regression, an ensemble approach, and random forest regression. The results demonstrated that ADePT generally outperformed other methods when predicting inpatient admissions and separations for multiple forecast horizons. For ED presentations, differences in accuracy were not statistically significant. Importantly, ADePT also showed high accuracy when applied to smaller patient subgroups, including emergency and elective inpatient admissions. By providing reliable short-term and long-term forecasts, ADePT could support more effective daily bed management as well as improved long-term capacity planning.

Clinical documentation burden remains a significant challenge in healthcare, particularly in primary care settings. Artificial intelligence (AI) scribes have emerged as potential solutions, but their effectiveness compared to human documentation lacks robust evidence, especially in community general practice environments. Documentation quality is compared between four commercial AI scribes and human-generated notes using four standardised clinical scenarios from the Royal Australian College of General Practitioners examination repository in simulated general practice consultations. Three experienced general practitioners, blinded to the source, assessed quality using a modified Physician Documentation Quality Instrument (PDQI-9). AI-generated notes outperformed human documentation across multiple quality domains. Top AI scribes scored a mean of 44.08/50 (SD = 3.32) vs. 37.42 (SD = 9.78) for humans, excelling in thoroughness (M = 4.92), accuracy (M = 4.67), and freedom from bias (M = 4.92). Inter-rater reliability was high for thoroughness (ICC = 0.879) and accuracy (ICC = 0.745), but lower for subjective areas like synthesis (ICC = 0.082). This study shows that AI scribes can outperform traditional documentation in simulated general practice. Successful implementation, however, depends on workflow integration and customisation. Standardised evaluation and balancing consistency with clinical context are key. Future research should explore real-world use, focusing on customisation and workflow impact.

The increasing use of digital technologies in healthcare in Australia necessitates the need for higher education institutions to equip undergraduate students with the necessary skills to use these digital tools effectively and efficiently. The rise in consumer engagement and expectations requires graduates to be more digitally literate to provide education to their clients. Equipping students with key foundational digital health skills and knowledge is important when students are entering the real world through practicums and eventual graduate programs.

This study investigates the persistent underperformance of health information systems (HISs) in rural Indonesian mental healthcare, despite national digital health initiatives. Utilising a socio-technical systems theoretical lens, an eight-month exploratory qualitative study was conducted, involving focus groups, in-depth interviews with healthcare providers, community health workers, and residents, alongside a literature review. Thematic analysis identified three critical socio-technical misalignments hindering HIS effectiveness: severe data integration issues due to fragmented tools and lack of interoperability; significant resource constraints (technical, human, and budgetary), and pervasive cultural and social stigma, which impede help-seeking, data accuracy and holistic care delivery. The study concludes that these are not technological failures but systemic design breakdowns, and calls for a situated, multi-stakeholder approach to co-design context-sensitive, user-centred HISs that integrate informal work systems, thereby laying foundations for equitable mental healthcare in resource-limited environments.

Cerebral palsy is the most common physical disability and the fifth most common cause of death in childhood. There is no known cure for this lifelong condition that has complex variations in symptoms and severity. Families are faced with challenges in how to find new, safe and effective interventions and how to choose treatments that align with their family priorities. People with the lived experience of cerebral palsy were connected with clinical experts, software developers and mHealth researchers through focus groups and workshops. Together, a mobile health (mHealth) aide was codesigned and developed to streamline and filter treatments based on family priorities. The aide contains a step-by-step guide, a search function, treatment factsheets, and support resources to empower evidence-based personalised decision making. The mHealth app has been endorsed by research partners and will be freely available in app stores worldwide.

Interoperability between blockchain platforms remains a key challenge, particularly in sensitive domains such as healthcare, where the secure and consistent exchange of clinical information between institutions is essential. While technical interoperability solutions exist, semantic interoperability at the level of smart contracts continues to be a significant limitation. This paper presents MUISCA, a mechanism based on Model-Driven Engineering that enables the automatic generation of interoperable smart contracts across different blockchain platforms. By defining metamodels, abstract models, and transformation rules, MUISCA produces platform-specific code for technologies such as Ethereum and Hyperledger Fabric. The mechanism was validated through a healthcare case study focused on patient transfers between medical institutions, demonstrating its ability to support the secure exchange of clinical data. Additionally, its acceptance was evaluated through expert surveys assessing perceived usefulness and ease of use. Results show that MUISCA improves smart contract portability, reduces implementation errors, and enhances system security. The proposed solution contributes to advancing semantic interoperability in blockchain-based health information systems and provides a foundation for broader application in other critical domains that require high levels of integration and data protection.

Poor medication adherence remains a persistent challenge in healthcare, significantly impacting treatment outcomes and healthcare costs. While reminders and education have shown limited success, recent developments in behavioral economics suggest that subtle interventions, known as "nudges", can influence patient behavior more effectively. This paper presents the design, development, and initial evaluation of a smartphone application aimed at improving medication adherence through nudging techniques and interactive features. The app combines behavioral design principles with human-centered development to offer functions such as context-aware reminders, a social avatar interface (Adii), symptom and appointment tracking, and customizable scheduling. Nudging strategies include default settings, motivational prompts, social reinforcement, and salience through feedback mechanisms. The app's structure was co-designed with healthcare stakeholders, informed by literature and market analysis, and implemented using React Native for cross-platform compatibility. A two-phase usability study with 16 participants revealed that default schedules and visual feedback significantly influenced adherence behaviors. Personalized reminders and the avatar enhanced emotional engagement, while onboarding ease and offline support improved user trust. Though still in prototype phase, the app demonstrates promising utility for long-term adherence improvement. Future versions aim to incorporate adaptive nudging based on AI-driven user behavior modeling.

This study examines the use of the Synthetic Data Vault (SDV) tool in generating synthetic EHR data for adverse drug events (ADE) detection. Experiments were conducted with three off-the-shelf synthetic data generators: GaussianCopula, Conditional Tabular Generative Adversarial Network (CTGAN) and Tabular Variational Autoencoder (TVAE), using a structured Swedish dataset. Evaluations included SynthEval metrics and downstream performance assessment using a 'train-on-synthetic, test-on-real' (TSTR) approach with Random Forest classifiers. Results show that TVAE's performance varied with dataset size and class balance, with larger datasets improving its performance. GaussianCopula provided more stable utility and stronger privacy protection at the cost of fidelity. CTGAN generated realistic data but exhibited inconsistent performance under TSTR evaluation. These findings highlight the importance of selecting synthetic data models based on healthcare application needs and dataset characteristics.