Background: Cardiovascular disease (CVD) remains one of the leading causes of mortality worldwide, and early detection of high-risk individuals is crucial for effective prevention and intervention. Traditional risk assessment models, while useful, often rely on limited data and may lack the precision needed to identify emerging risk patterns. This study explores the application of machine learning techniques to enhance the identification of individuals at high risk of developing CVD, using a broader set of health indicators and real-world data.

Methods: We developed a machine learning model that analyzed extensive data from electronic health records, lifestyle factors, and demographic information from a large patient cohort. Key features included blood pressure, cholesterol levels, physical activity, and socioeconomic factors, integrated with machine learning algorithms to predict future CVD risk with higher accuracy than conventional models. Model validation was conducted using a separate test dataset to ensure robustness and reliability.

Results: The machine learning model significantly improved risk prediction accuracy, achieving a 20% increase in sensitivity compared to traditional methods. This improvement allowed for more targeted recommendations for lifestyle changes, screenings, and preventive interventions. Furthermore, the model identified novel risk patterns, such as the combined impact of socioeconomic and lifestyle factors, which had previously been underrepresented in standard assessments. These insights have the potential to refine public health strategies for CVD prevention.

Conclusion: Integrating machine learning into CVD risk assessment provides a powerful tool for early intervention, allowing healthcare providers to identify high-risk individuals with greater precision. This approach not only improves individual outcomes but also supports more efficient resource allocation in healthcare settings. Further research is recommended to explore the scalability of this model across different populations and to investigate its long-term impact on CVD prevention strategies.

Background: Artificial intelligence (AI) is revolutionizing epidemiology by enabling real-time surveillance and predictive modeling, essential tools for rapid infectious disease control. Traditional epidemiological methods, while effective, often struggle to deliver timely insights needed for proactive outbreak detection and efficient public health interventions. This study explores integrating AI-driven surveillance systems to enhance the speed and accuracy of infectious disease monitoring and response.

Methods: An AI-based surveillance model was developed using machine learning algorithms to analyze diverse data sources, including electronic health records, social media trends, and environmental sensor data. The model was implemented in partnership with public health agencies across multiple regions to evaluate its effectiveness in early outbreak detection and prediction. Continuous learning enables the model to dynamically adapt to new data patterns, increasing its predictive accuracy and reliability over time.

Results: The AI-driven model demonstrated a significant reduction in detection lag, identifying outbreak clusters 30% faster than traditional methods. Early detection allowed timely public health interventions, effectively limiting infection spread in various regions. Additionally, the system provided public health agencies with improved operational efficiency and optimized resource allocation based on data-driven insights. Data privacy was rigorously upheld through robust anonymization protocols, ensuring ethical compliance and safeguarding individual information.

Conclusion: Integrating AI into infectious disease surveillance represents a transformative advancement for public health, offering a faster, more accurate approach to outbreak monitoring and control. This approach reflects the future of epidemiology, embracing digital innovation to strengthen surveillance capabilities. AI-driven models facilitate quicker, data-informed interventions, contributing to more effective infectious disease control. Further research is recommended to expand AI applications across additional epidemiological fields and to develop advanced data privacy measures to address ethical considerations in real-time surveillance.

Background
Excessive sugar consumption, especially from sugar-sweetened beverages and ultra-processed foods, is a major public health concern, contributing to the burden of non-communicable diseases. Additionally, social inequality in sugar consumption is evident, with individuals from lower socio-economic status (SES) groups typically consuming more sugar. Front-of-Package Labeling (FOPL) is a population-level policy tool recommended by World Health Organization to improve diet and reduce diet-related health burden. FOPL may also reduce social inequality in sugar consumption. While systematic reviews have examined the impact of FOPL on sugar purchasing and consumption, limited research has assessed its effectiveness from an inequality perspective. To fill this gap, this systematic review examines the impact of FOPL on social inequalities in actual sugar consumption and purchasing behaviors.
Methods
A systematic search was conducted across four databases. Eligible studies included experimental, observational, and simulation designs, with no restrictions on age or geography. Studies which report on the impact of FOPL on actual sugar consumption or purchasing across SES with a comparison group without FOPL were included.
Results
Eleven studies met the inclusion criteria, primarily natural experiments (n=6). FOPL formats were examined, including health warning labels (n=8), traffic light systems (n=2), and Keyhole (n=1). All studies were conducted in high- and upper-middle-income countries. All studies reported that FOPL significantly reduced sugar consumption. Eight studies suggested that FOPL may reduce social inequality in sugar consumption.
Conclusion
Overall, FOPL reduced sugar consumption at the population level, with some evidence indicating a potential to narrow social inequality. However, further research is needed to better understand the interaction between FOPL effectiveness and SES and identify the modifiers influencing sugar consumption to develop strategies for optimizing FOPL to reach disadvantaged populations.

Background
While a target has recently been set in the Australian Government’s National Tobacco Strategy to reduce daily smoking rates to ≤5% by 2030, this fails to explicitly address existing smoking-related inequity across sociodemographic groups. We aimed to quantify the future health impact of three hypothetical scenarios: 1. Immediate eradication of tobacco use, 2. Achievement of the Government target (but with remaining inequity), 3. an equity scenario where all remoteness by SEIFA by Indigenous status strata achieve the 2030 target.
Methods
A Markov process was constructed to simulate future smoking behaviours in the Australian population, with a proportional multistate lifetable (PMSLT) that sums the health impacts of 31 smoking-related diseases. The model outputs the difference in deaths and health-adjusted life years (HALYs) for the Australian population under each hypothetical endgame scenario in comparison to ‘business-as-usual’ (BAU), over 20-40 years.
Results
Complete eradication of tobacco smoking in the Australian population, compared to expected BAU trends, could result in a gain of 682,000 (95% UI 525,000-890,000) HALYs, and avert 42,300 (95% UI 34,050-51,800) deaths, from 2024-2044. Achieving the Government target would result in approximately 39% of the possible benefit, compared to 49% under the equity scenario. The relative gain under the equity scenario is greater for more disadvantaged strata. Under the equity scenario in comparison to BAU, the all-cause mortality gap between the most and least deprived SEIFA quintiles decreased by 5% at the population level by 2064.
Conclusions
Moving beyond an overall Government target to an equity target that more proactively closes gaps in smoking by social groups by 2030 will not only increase overall health gain but also contribute to meaningful reductions in health inequalities in Australia. Future research should focus on the health gains across scenarios for the First Nations population, with appropriate leadership by Indigenous researchers.

Title: Identifying medicine shortages: novel epidemiologic approaches and a case study for ADHD

Background
Lisdexamfetamine is first-line treatment for attention deficit hyperactivity disorder (ADHD). In December 2023, the Therapeutic Goods Administration (TGA) was notified of limited availability of lisdexamfetamine due to manufacturing issues. Dispensing claims data are publicly available and can be used to monitor medicine use at the population level, however, the use of these data to explore the impact of medicines shortages is limited.

Aim: To develop a novel approach to monitoring medicine dispensing data to determine the impacts of shortages on medicines use and access in Australia.

Methods
Publicly available Pharmaceutical Benefit Scheme Section 85 monthly supply data (January 2022–August 2024) were analysed. Dispensed prescriptions and monthly percentage changes were calculated. Defined Daily Doses (DDD) per 1000 population were used to measure medicine consumption at the population level over time.

Results
Lisdexamfetamine dispensings increased by 170% from January 2022. The 30mg shortage (Aug 2023–Mar 2024) led to a 68% drop in dispensings from August to September 2023, while the 50mg shortage (Nov 2023–Mar 2024) caused an 80.5% decline in dispensings. During these shortages, 40mg dispensings increased by 49%, eventually experiencing its own shortage due to increased demand. DDD analysis showed more than 50% reductions for 30mg and 50mg, while the 40mg DDD nearly doubled. No significant changes in DDD were observed for 20mg and 70mg strengths during the shortage period.

Conclusion
Lisdexamfetamine shortages lead to disruption of the market. While access to treatment appeared to be unaffected overall, there appeared to be a substantial change in the specific product strengths people were using. Identification of market disruptions using novel close-to-real-time monitoring of dispensing patterns and fluctuations during shortages can help to mitigate threats to access of medicines and identify areas for further research.

Background: Across the world, Indigenous Peoples participation in population-based cancer screening programs are consistently lower than their non-Indigenous counterparts. Financial and other incentives have been used to increase participation in cancer screening, however, their effectiveness among Indigenous populations remains unclear. Hence, we conducted a systematic review to investigate the effectiveness of incentives on cancer screening uptake among Indigenous populations.

Methods: This systematic review adhered to PRISMA guidelines. Four databases (PubMed, Web of Science, EMBASE, and CINAHL) were searched for published studies from inception to June 2024. Eligible studies included qualitative, quantitative, or mixed methods, those reporting on breast, colorectal, lung and cervical cancer screening participation outcomes. Two independent reviewers, screened titles, abstracts, and full texts, followed by data extraction and quality appraisal using the Mixed-Methods Appraisal Tool.

Results: The 35 included studies reported on education programs (65.7%), an outreach program (22.9%) and 11.4% used a combination of both. Education-based interventions significantly improved cancer screening participation rate (11.7% to 370%) and/or knowledge among Indigenous populations. Outreach programs, particularly those with personalized follow-ups, boosted participation by over 31%. Combining outreach and education also proved effective, especially with trained healthcare workers and tailored community approaches. Financial and non-financial incentives like transportation and childcare increased engagement. Studies co-designed with Indigenous communities, were also more effective in improving knowledge, attitudes, and increasing screening rates. While few studies showed no significant results, most demonstrated improved screening behaviours and attitudes, emphasizing the effectiveness of incentives.

Conclusion: Our findings underscore the critical role of Indigenous co-design in developing and implementing incentives such as educational and outreach programs, financial support, and transportation assistance to increasing screening participation rates among Indigenous populations. This evidence should be considered while formulating cancer screening policies and programs and could assist in closing the cancer screening gap in Australia for First Nations Peoples.

Background
Global air travel plays a major role in the translocation of infectious diseases. Emerging infectious disease threats significantly impacts air travel, necessitating effective passenger screening strategies to mitigate the translocation risk, particular in pandemic events. This study evaluates various screening methods, including pre-departure and post-arrival testing, quarantine compliance, their effectiveness in identifying infectious passengers, and a framework to apply evidence-based screening to other emerging infectious disease threats.
Methods
We employed adapted Monte Carlo simulations to model hypothetical disease timelines for traveling passengers. Key screening strategy factors assessed included the number and type of RT-PCR and antigen tests administered before and after travel, as well as quarantine length and compliance. Data on COVID-19 infections and population demographics were sourced from the Johns Hopkins University repository, and model calibration was performed using real-world data from Iceland and Canada.
Results
Our findings indicate that post-arrival testing and high quarantine compliance are the most critical factors in reducing the translocation of COVID-19. The analysis revealed that combining quarantine with post-arrival testing can effectively shorten quarantine durations while maintaining screening sensitivity. Notably, the timing between tests significantly influenced sensitivity, with longer intervals between tests yielding better results. Among the screening strategies evaluated, dual RT-PCR tests demonstrated the highest sensitivity, while dual antigen tests showed the most variability.
Conclusion
This study underscores the importance of tailored screening strategies in air travel to effectively identify infectious individuals to reduce disease translocation while maintaining travel between regions. Our findings highlight the use of rapid antigen testing as a viable alternative to RT-PCR, offering a less burdensome approach to passenger screening. The flexible framework developed here is adaptable to other existing or emerging diseases and can inform public health policy and enhance pandemic preparedness for future infectious disease outbreaks tailoring screening to minimise translocation risk.

Background: Children and adolescents are exposed increasingly to processed food in China, however, its association with obesity has not been investigated. We aimed to assess the consumption of ultra-processed food (UPF) and its association with overweight/obesity among children and adolescents in China.
Methods: 3,437 children and adolescents aged 6-18 years, participating at least twice in the China Nutrition and Health Survey were included. Food intake was collected using a 3-day 24-hour dietary recall method at home visits. Body weight, height, and waist circumference (WC) were measured during the survey. UPF was defined by food process levels using NOVA classification. Overweight/obesity was defined by the international age- sex-specific BMI and WC cut-offs. The association between UPF consumption and overweight/obesity were assessed using mixed effect logistic regression analyses adjusted for sociodemographic, economic, behavioural, dietary, and health factors.
Results: The mean daily UPF consumption of the study population (mean age 9.3 years) increased from 9.7 in 1997 to 60.0 grams in 2011. The adjusted odds ratios (OR) (95% CI) for overweight/obesity (using BMI) for UPF consumption of 0, 1-49, 50-99, and >100 g/day were 1.00, 1.38 (0.98-1.94), 2.01 (1.25-3.24), and 1.53 (0.82-2.86), respectively (P-trend =0.013). Similarly, the corresponding adjusted ORs (95% CI) for central obesity (using WC) were 1.00, 1.84 (1.30-2.60), 2.13 (1.30–3.48), and 2.15 (1.14–4.05) (P-trend<0.001).
Conclusions: Higher long-term UPF consumption was associated with an increased risk of overweight/obesity among children and adolescents in China.

Introduction
Risk factors for respiratory tract infections in patients on mechanical ventilation include prolonged stay in the ICU and colonisation of the lower respiratory tract by oropharyngeal microorganisms. In the ICU of a quaternary facility, oral care guidelines were developed to standardise mouth care for critically ill patients on mechanical ventilation and reduce the risk of respiratory complications. This study aimed to assess overall adherence to the work-based guidelines to enhance service quality.
Methods
A cross-sectional review was undertaken of records on the computer information system (Metavision iMDsoft ) for adult patients 18 years or older admitted to the ICU between January 1st, 2021, and June 31st, 2021. Patients selected were critically ill adults aged 18 years or more who were receiving mechanical ventilation. Univariate and bivariate analysis were undertaken to explore adherence to oral guidelines and associations between oral care and patients’ clinical characteristics. Data analysis was performed using StataCorp. 2023. Stata Statistical Software: Release 18. College Station, TX: StataCorp LLC.
Results
Of the 425 patients included, 264 (62.1%) were male, and their mean age was 58.4 years. The mean length of stay in the ICU was 6.4 days, with 49.80% of patients intubated 48 hours or more. Oral care was provided to most of the patients (n=414; 97.4%), but the type of care provided varied daily. The most common oral care practice done per person on the first day was paraffin to the lip (4.10), followed in descending order by water (3,72), chlorhexidine (1.98), sodium bicarbonate (1.68), toothbrush (0.73), artificial saliva (0.37) and the use of toothpaste. A similar pattern was observed on average per day per length of stay.

Conclusion
The provision of oral hygiene care varied widely between patients. Predominantly, oral moisturising care was observed, with less focus on the mechanical removal of plaque.

Background: Features of neighbourhoods, including green space, have been shown to shape obesity and health in adults. However, research is predominantly from Europe and North America.

Purpose: We explored the cross-sectional association of geospatially-derived indicators of neighbourhood greenness, tree cover and built-up areas with overweight and obesity in Bangladesh. We tested whether associations varied by urbanicity and household income.

Methods: We used data from 42,081 Bangladeshi adults enrolled in South Asia Biobank between 2017-2022. Using a geographical information system, we constructed a 500m buffer around home address. We quantified greenness using the normalized difference vegetation index. We quantified tree cover and built-up areas using the WorldCover dataset. Adjusted regression models explored the association of (a) greenness; (b) tree cover and (c) built up areas with overweight and obesity defined using (i) body mass index; (ii) waist circumference and (iii) waist-hip ratio. We tested for interactions by urbanicity and household income.

Results: 49% of participants were overweight or obese. Higher greenness was associated with lower odds of overweight and obesity across body mass index (OR=0.90, 95%CI 0.86 to 0.93), waist circumference (OR=0.87, 95%CI 0.84 to 0.90) and waist-hip ratio (OR=0.87, 95%CI 0.84 to 0.91). Conversely, higher amounts of built-up areas were associated with higher odds of overweight and obesity across body mass index (OR=1.10, 95%CI 1.08 to 1.11), waist circumference (OR=1.08, 95%CI 1.07 to 1.10) and waist-hip ratio (OR=0.87, 95%CI 0.84 to 0.91). Associations were strongest in urban participants. Tree cover associations were not statistically significant.

Conclusions: In Bangladeshi adults, higher greenness and lower amounts of built-up areas were associated with more favourable body composition, particularly in urban dwellers.

Practical implications: This study aligns with others indicating beneficial associations of green space with health, and with policy to support greening of cities.

Funding: National Institute for Health Research (16/136/68, 133205).

Causal inference in environment-health research is frequently limited by a lack of longitudinal exposure data linked to longitudinal health data. Therefore GenV, an ongoing mega-cohort of >120,000 children and their parents, aims to address this limitation by creating a suite of longitudinal environmental exposures at high spatio-temporal resolution. Here we report on version 1 of this suite, which covers the GenV baby preconception, pregnancy and birth windows.

Methods: We sourced spatial data from satellite images, open street maps, and the Australian Bureau of Statistics and used geospatial methods to create indicators of air pollution, greenness, land surface temperature, food environments, and disadvantage for 2019-2023. To overcome challenges with accessing participant residential data we calculated these indicators for all addresses in the state of Victoria.

Results: We generated a suite of environmental indicators for every address in Victoria and covering the period from 2019-2023. These data show distinct spatio-temporal patterns, and will be linked to the GenV data, and in the future other linked datasets (e.g. perinatal, hospitalisations, ABS). We are in the process of adding different environmental indicators.

Conclusion: The suite of high spatio-temporal environmental indicators linked to GenV data will form part of the GenV Open Science Resource. It will enable Australian researchers to address a range of environmental-health research questions, model environmental interventions for improved health, and identify areas which would derive the most health benefit from healthier environments.

Background: Diet influences MS risk and progression, with strong evidence for the Mediterranean diet, but mechanisms remain unclear. We recently found that differential DNA methylation (DNAm) modules mediated the associations of environmental and genetic risk factors with MS. This study examined whether DNAm modules mediate the associations between alternative Mediterranean diet score (aMed) and alternative Mediterranean diet indices (aMed-RED) with MS risk in the Ausimmune Study.
Methods: We analysed data from 206 cases and 347 matched controls. Diet intake was assessed using the Cancer Council Victoria food frequency questionnaire. The molecular mediation analysis estimated the total, direct, and indirect effect, and proportion mediated by the mediator in the association between diet and MS. Proportion of total effect acting via DNAm modules was reported as the proportion mediated (indirect/total effect ratio).
Results: The aMedRED (aOR=0.88; 95% CI: 0.79, 0.97; p = 0.014) and healthy diet score (aOR=0.87; 95% CI: 0.77, 0.99; p = 0.048) were significantly associated with reduced risk of MS risk onset.
We showed that the Yellow DNAm module significantly mediated associations of aMED (30.3%, p=0.037) and aMED-Red (22.1%, p=0.027) with MS risk. The proportions mediated by Yellow DNAm for the association between aMed and aMedRED with MS onset-risk was 30.3% and 22.1%, respectively.
Stratifying by HLA-A:02 status, the Yellow DNAm module significantly mediated the association of the aMedRED score and MS risk onset among those positive for HLA-A:02 genotype (aOR=0.95, p=0.019) and percent mediated was 45%. Additionally, high aMedRED score and HLA-DRB1*15 risk genotype was positively and significantly associated with MS risk compared to neither factors (Synergy index=1.41, p=0.008).
Conclusion: Up to a third of the effect of Mediterranean diet on MS risk act via differential DNAm in biologically plausible genes related to immune and neurological function. These findings inform DNAm-based screening and pro/de-methylation interventions in MS.

Background: Circumstances surrounding clinical decisions about the medical suitability and safety of deceased organ donors for kidney transplants can vary. We sought to estimate the maximum gains achievable from interventions that increase clinical certainty and consistency in deceased organ donor medical suitability decisions, potentially increasing organ donation and transplant opportunities. Methods: Using a data-linked biovigilance register, we described the clinical characteristics of potential kidney donors and categorised the circumstances where potential donors were deemed not medically suitable for transplantation in NSW, Australia, 2015-2022. We then estimated the potential maximum number of donors per million population that could have been gained if these donors had been accepted. Results: Of 5,211 potential donors deemed not medically suitable, 674 (13%) were acceptable using current Australian and New Zealand clinical guidelines. On average, 84 (1.6%) donors were declined each year due to risk aversion (higher but acceptable risk of cancer and/or infection); accepting these donors would increase annual donation rates to 27.4 donors per million population. A further 1,660 (31%) had insufficient clinical information at the time to assess medical suitability accurately, of whom 226 (14%) lacked sufficient pathological and treatment details of their confirmed cancer diagnoses, 538 (32%) had increased risk behaviours for a blood-borne virus but were not sent for testing, and 674 (41%) required more information about current infection control and treatment. The proportion with insufficient clinical information dramatically decreased in 2016 after the introduction of real-time access to centralised cancer registrations but slowly increased from 2018 after revision of clinical guidelines (73% in 2016, 9% in 2018, 21% in 2021). Conclusion: Accepting the potential donors forgone as not medically suitable but within current clinical guidelines could significantly increase donation rates in NSW, Australia. Interventions to address clinical risk aversion and to increase real-time access to granular clinical data are warranted.

Background:
Mediterranean diet adherence is cross-sectionally associated with better mental health and quality of life (QoL) in people living with multiple sclerosis (plwMS), but prospective associations remain unexplored. To investigate whether greater Mediterranean diet adherence predicts long-term mental health and QoL outcomes in plwMS.
Methods:
Data were obtained from plwMS in the UK MS Register at baseline and the 6-year follow-up. We examined whether Mediterranean diet adherence, assessed using the alternate Mediterranean score (aMED; 0–9), at baseline predicted mental health and QoL outcomes at 6-years: anxiety and depression (Hospital Anxiety and Depression Scale), psychological impact of MS (MSIS-29 Psychological Subscale), and QoL (EuroQoL Visual Analog Scale). Quantile and log-binomial models were adjusted for relevant demographic/clinical covariates.
Results:
Higher baseline aMED was prospectively associated with lower anxiety (adjusted co-efficient, aβ=-0.17; 95%CI=-0.34,-0.01), reduced symptoms of depression (adjusted relative risk (aRR)=0.93; 95%CI =0.87,0.99), lower psychological impact of MS (aRR=0.94; 95%CI=0.88,0.99), and higher QoL (aβ=1.25; 95%CI =0.18,2.33). Participants in the highest aMED tertile (aMED≥6) had 33% reduced risk of higher psychological impact of MS (aRR=0.67; 95%CI=0.48,0.95) and significantly higher QoL (aβ=6.7; 95%CI=1.32,11.98).
Conclusion:
Greater Mediterranean diet adherence predicts better mental health and QoL six years later, suggesting potential long-term benefits for plwMS. With validation in future studies, the Mediterranean diet could become a practical and evidence-based dietary recommendation as part of a comprehensive chronic disease management plan, complementing established clinician-directed care.

BACKGROUND
People diagnosed with multiple sclerosis (MS) often seek to make dietary changes. Currently, they are advised to follow national dietary guidelines. A tool to measure adherence with dietary guidelines in people with MS is required.
OBJECTIVES
Develop a tool to measure adherence with the Australian Dietary Guidelines (ADG) in a cohort of Australians living with MS.
METHODS
Dietary intake data in the Ausimmune Longitudinal (AusLong) study is collected using the Dietary Questionnaire for Epidemiological Studies Version 2. The ADG database was previously developed to evaluate eating patterns in the 2011-13 Australian Health Survey. In this study, an ADG adherence tool (ADG-AT) was developed by systematically matching AusLong and ADG food items by food descriptions. The final ‘match’ resulted from either a single, direct match or an average of multiple ADG matches. The ADG-AT was then applied to AusLong dietary data at baseline for participants with MS having plausible energy intakes (3,000-20,000kJ/day) to estimate consumption of the five food groups (to consume) and discretionary foods (to limit). Adherence with dietary recommendations was evaluated according to age and sex, where participants within 0.5 serves of the recommendation were considered adherent.
RESULTS
Applying exclusion criteria resulted in 131 participants. For each food group and discretionary foods, the proportion of participants meeting recommendations and the median (IQR) number of serves consumed were: vegetables 5.6%, 2.3 (1.8-3.0) serves; fruit 50.0%, 1.5 (0.9-2.0) serves; grains 30.0%, 4.2 (3.5-5.3) serves; meat/alternatives 56.7%, 2.2 (1.6-2.8) serves; dairy/alternatives 44.4%, 2.0 (1.5-2.5) serves; and discretionary foods 35.6% 3.7 (2.5-4.9) serves.
CONCLUSIONS
The ADG-AT enabled food group intake analysis and ADG adherence evaluation in a cohort of Australians living with MS. The tool can be used to further explore dietary associations with MS outcomes and may be adapted in other studies using different data collection methods and/or food composition databases.