Background and Aim

Although artificial intelligence (AI) is widely used due to its ability to empower educators and learners, the rapid advancement of generative AI (GenAI) has triggered a global debate on how to regulate this new technology in teaching and learning activities. This study aimed to explore challenges and best practices for promoting responsible AI use among public health trainees at Fiji National University, thereby fostering an ethical and productive learning environment.


Methods and Analysis

This study used qualitative methods, including in-depth interviews and document analysis. A total of 15 interviews were conducted among public health students who used AI. Additionally, 10 students who used AI to complete their assignments were asked to redo them without AI assistance to compare quality, accuracy, and consistency using a standardised marking rubric. Relevant documents and global AI regulatory frameworks were also reviewed to provide context. Data were analysed using a thematic analysis approach.


Outcome

Students in the study area used AI for personalised learning, generating supportive materials that enabled them to study at their own pace. Overdependence on AI stemmed form easy information access and a lack of clear educational guidelines. Students also expressed uncertainty regarding ethical boundaries, noting AI’s potential to undermine academic writing skills and critical thinking. To promote responsible AI use, best practices must include enforcing clear AI guidelines, providing targeted educator feedback tailored to individual student performance, and designing assessments methods that emphasize originality and analytical thought.

Conclusion and Future actions

While AI tools have the potential to enhance learning, clear policies are essential to safeguard academic integrity and critical thinking. Higher education institutions must implement structured guidelines, targeted training, and alternative assessments to ensure students use AI responsibly.

Background and Aim:
Artificial intelligence (AI) holds significant promise for health systems facing resource constraints, geographic dispersion, and workforce shortage acutely felt across the Pacific. Yet little is known about whether Pacific public health workers (PHW) have the knowledge, access, or infrastructure to harness this potential. This cross-sectional baseline survey aimed to characterise the current state of AI awareness, training, use, and readiness PHW across Pacific Island countries and territories (PICTs).

Methods and Analysis:
An anonymised online survey was conducted in 2025 across PICTs through the Pacific Public Health Surveillance Network (PPHSN). Responses were collected from 253 PHW across 20 PICTs. The survey captured respondents' professional roles, AI familiarity, training history, current use, governance and infrastructure context, perceived challenges, and attitudes toward AI investment. Descriptive analysis was undertaken across all domains.

Outcomes:
Overall, 77% (n=197) of respondents reported low to moderate AI familiarity and 88% (n=203) had received no formal or informal AI training. While 16% reported regular AI use and 57% occasional use, 11% had never used AI tools. Structural barriers were prominent: 55% reported no dedicated AI funding within their health sector, 67% reported no AI governance frameworks, and 66% considered their country only somewhat or not equipped for AI adoption. Despite these gaps, respondents expressed optimism, identifying AI’s potential to improve data management, surveillance and service delivery in remote, workforce-constrained settings.

Conclusion and Future actions:
PHW across the Pacific recognise AI as a promising public health tool, but largely without foundational safeguards. This disconnects risks constraining benefits and amplifying harms, such as unsafe use, inequitable access, or widening digital divides. Addressing these gaps through targeted AI literacy programmes, fit-for-purpose governance, and equitable digital investment is essential to maximise benefits while minimising risks. Given its regional mandate and networks, SPC is well positioned to support coordinated capacity-building and guidance for responsible, context-appropriate AI use across Pacific health systems.

Background and Aim
Artificial intelligence (AI) is increasingly being integrated into workplaces to support communication, decision-making, and data analysis. While AI has been widely explored in clinical healthcare settings, little is known about its use within the Australian public health and health promotion (PHHP) workforce. This study aimed to identify the extent of AI use, the tools being utilised, workforce perceptions, and patterns of adoption across the PHHP profession.

Methods and Analysis
A national cross-sectional online survey was conducted with Australian PHHP professionals. The survey examined AI use, perceived usefulness, workplace readiness, and organisational support. Data were analysed using descriptive and inferential statistics in SPSS.

Outcomes
Preliminary analysis suggests a range professionals, including health promotion practitioners, policy officers/analysts, health program coordinators, epidemiologists, and other PHHP roles used AI in their work within the previous 12 months. Generative AI was the most used technology, with ChatGPT and Microsoft Copilot being the primary tools. AI was commonly used for report drafting, communication activities, and administrative tasks with many participants reporting daily or frequent use (2-4 times per week). AI adoption occurred through both employer-endorsed and personal use, and most respondents agreed that AI enhanced their work effectiveness.

Conclusion and Future Actions
AI is already widely used across the Australian PHHP workforce, particularly through generative AI tools. Commonly reported applications related to drafting, communication, and administrative tasks, suggesting AI is currently being used for general workplace productivity. This raises questions about the extent to which AI is being applied to activities specific to PHHP practice and highlights the need to better understand how professionals are integrating these technologies into their work. Future research should explore factors influencing AI adoption, the opportunities and challenges with its use, and required knowledge, skills, and ethical capabilities. These insights will inform workforce development and higher education curricula.

Background and Aim
Health misinformation is a growing public health challenge, with AI-generated content accelerating its spread across digital platforms. As heavy users of these platforms, young people need the skills to recognise how this misinformation spreads and to critically appraise what they encounter. This study addresses this gap by co-designing an innovative, engaging online game to build AI health literacy skills among Australian high school students.

Methods and Analysis
An iterative co-design process was conducted with students in Year 9-10 (aged 14-17 years) and their teachers. The process included multiple rounds of small-group workshops to develop engaging and relevant lesson content aligned with teaching curriculum. Content writers and game designers developed workshop ideas into a prototype version of the digital game. User testing of the prototype was also conducted in subsequent workshops for further participant feedback to refine in-game features and user experience.

Outcomes
More than 20 students and teachers participated in the co-design process. Workshops highlighted the need for the game to build practical skills to identify and counter AI-driven health misinformation, using health scenarios aligned with student interests and curriculum priorities. The final design features a competitive, interactive format allowing students to play against each other, with national rollout planned during Health Literacy Month in October.

Conclusion and future actions
As AI reshapes the digital information environment, adolescents need targeted skills to navigate AI-generated health content. Co-design can ensure educational interventions are relevant and engaging for students, as well as acceptable for their teachers. Forthcoming evaluation will test the effectiveness of the game in improving AI health literacy in students and inform future scalable approaches.

Background and Aim: Artificial intelligence (AI) systems can process large volumes of data generated in medicine. Developing countries such as Pakistan are lagging in implementation of AI-based solutions. This study explored perceptions and readiness of healthcare professionals toward AI in medicine and their satisfaction with chatbot-generated responses.
Methods and Analysis: A cross-sectional study was conducted among healthcare professionals working in public and private hospitals, educational institutions, and community pharmacies in Faisalabad, Pakistan, from October 2023 to March 2024. Convenience sampling was used. Data were collected using a validated questionnaire comprising four sections: demographics, perception of AI, readiness for AI adoption, and satisfaction with chatbot responses. Participants interacted with the Poe (Platform for Open Exploration) chatbot using self-generated queries under controlled conditions. Chatbot responses were evaluated for satisfaction. Descriptive statistics and chi-square tests were applied.
Outcome: A total of 353 healthcare professionals participated, including pharmacists (n=136), doctors (n=110), and nurses (n=107). Although attitudes toward AI were generally positive, concerns were reported regarding its inability to be applied to every patient (81.8%) and its limited consideration of emotional well-being (82.4%). High agreement was observed for AI use in administrative tasks (74.8%) and research (83.9%). Readiness was highest for AI-based appointment scheduling (81.3%) and responding to health-related queries (78.2%). Overall, 77.3% reported satisfaction with chatbot-generated responses. Significant differences in satisfaction were observed among doctors, pharmacists, and nurses (p<0.05).
Conclusions and future actions: Healthcare professionals demonstrated generally positive perceptions and readiness toward AI, with favorable satisfaction regarding chatbot-generated responses. However, addressing concerns and providing necessary training are crucial for effective implementation.

Background and Aim: Generative AI (Gen-AI) is rapidly transforming health professions education, yet debate centres on its dual potential to enhance teaching while introducing risks to accuracy, equity, and academic integrity. While student use is well documented, the educator perspective remains underexplored. We aimed to explore educators' experiences, perceptions, and institutional support needs regarding Gen-AI integration.
Methods and Analysis: A convergent parallel mixed-methods study was conducted at an Australian university. Pre-course (n=6) and post-course (n=15) online surveys captured educators' Gen-AI proficiency, usage, concerns, and future intentions, and seven educators participated in semi-structured interviews. Descriptive statistics summarised survey data, and qualitative data underwent thematic analysis following Braun and Clarke's approach.
Outcomes: Educators translated their concerns into concrete changes to teaching and assessment. To safeguard academic integrity, some shifted to oral vivas in place of written assessment, designed exam questions around recently published evidence that Gen-AI could not yet access, and shared comprehensive lecture scripts and code to reduce student reliance on Gen-AI. Educators' self-rated proficiency increased: all reached at least beginner level post-course (up from 83.3%), with intermediate proficiency rising from 33.3% to 53.3%. Gen-AI use for teaching rose from 33.3% to 60.0% (predominantly ChatGPT), and familiarity with institutional guidelines doubled (33.3% to 66.7%). A "productivity paradox" emerged: time saved was offset by verifying outputs, limiting workload reduction. Concerns about data privacy (53.3%), bias (46.7%), and ethical implications (40.0%) were common.
Conclusion and Future actions: Educators viewed Gen-AI integration as inevitable but felt inadequately equipped. Institutions should provide sustained, pedagogically grounded training linked to academic integrity frameworks, clear policies distinguishing appropriate from inappropriate use, and ongoing support. Critically, integration strategies must account for the hidden costs of verification to safeguard accuracy where patient safety is at stake.

Background and Aim
The rapid integration of Artificial Intelligence (AI) in healthcare presents unprecedented opportunities to enhance public health outcomes, yet the health workforce remains largely untrained and unprepared. There is currently no unified strategy to build an AI-ready workforce at scale, with education pathways remaining fragmented across undergraduate and professional career continuums. Addressing this AI literacy gap is critical to ensuring that AI deployment in public health maximises benefits while actively minimising potential harms and inequities in care delivery.

Methods and Analysis
To address this gap, we are conducting a comprehensive desktop review as part of the DIGI-HEAL (Digital Health, Education, and AI Learning) initiative, a collaborative Australia-UK partnership. This review systematically evaluates how undergraduate health profession students and public health professionals in Australia, Scotland, and England are currently supported to deliver AI-enabled care. By synthesising existing literature, policy documents, and programmatic approaches across these major health systems, we are assessing the current landscape of AI training and education to inform the co-development of a best-practice capability framework.

Outcomes
Preliminary findings indicate that while isolated initiatives exist to improve AI literacy, they often lack a standardised structure, evaluation, and struggle with cross-system interoperability. Full analysis will be completed prior to the conference, detailing the specific approaches utilised across the participating countries. We anticipate our final results will highlight critical success factors for equitable AI education and identify specific gaps where current training pathways fail to adequately prepare the workforce for digital health innovation.

Conclusion and Future actions
The findings from this review underscore the necessity for robust, cross-institutional frameworks that govern AI capability-building in public health. Key lessons emphasise that successful workforce preparation requires rigorous co-design, multidisciplinary collaboration, and aligned international strategies. These insights will directly inform the nation-wide DIGI-HEAL Framework, offering actionable recommendations for educators and policymakers to responsibly integrate AI literacy into health professional training.

Background and Aim
The rapid integration of artificial intelligence (AI) in healthcare necessitates a generation of physicians capable of utilising and interpreting these advanced technologies. Consequently, there is a critical need to introduce medical data science early into medical education. This study aims to design, implement, and evaluate a newly piloted medical data science module for undergraduate medical students.

Methods and Analysis
An interdisciplinary curriculum was co-designed by lecturers from undergraduate medicine, postgraduate medicine, and medical technology. The module focuses on establishing foundational knowledge and practical applications of medical data science, intentionally emphasizing a “no-coding” approach tailored for medical students. To assess the module’s efficacy, an evaluation survey was deployed to measure student perceptions, attitudes, and knowledge self-assessment after the module.

Outcomes
This pilot was conducted at the Faculty of Medicine and Health of Sepuluh Nopember Institute of Technology, a pioneer of digital health education in Indonesia. The curriculum was successfully piloted with 48 third-year undergraduate medical students over a single semester. Student performance and comprehension were measured through a mid-term multiple-choice examination focusing on theory, critical appraisal via an AI journal reading assignment, and a collaborative final project. Working in groups, students practically re-implemented machine learning models on different datasets of clinical cases using the Orange data mining platform. They successfully executed data preprocessing, exploratory data analysis, and model evaluation metrics without writing code, culminating in a final presentation of their results.

Conclusion and Future actions
Initial evaluations indicate good overall student motivation and an adequate level of satisfaction with the learning media and curriculum design. However, student feedback suggested a strong desire for more practical applications. Future iterations of the module will address this by incorporating a higher volume of hands-on exercises, ensuring students are better equipped for modern clinical practice.

BACKGROUND AND AIM
The rapid adoption of generative AI tools has created a new and growing risk within public health organisations known as "shadow AI".

Publicly available and inexpensive ChatGPT, Copilot, Claude and other AI apps are collectively termed "shadow AI" if employees are informally using them outside official approval.

Shadow AI applications used in this way escape governance oversight, privacy assessment and risk management controls, potentially leading to a serious AI, data or privacy breach.

In healthcare, recent surveys indicate that nearly 20% of professionals personally admit to using unauthorised shadow AI tools at work.

Management traditionally respond to this risk by focusing on technical restrictions and IT controls. These measures have proven insufficient when faced with strong productivity incentives and widespread availability of shadow AI tools.

The aim of this presentation is to demonstrate why Shadow AI is primarily a people and culture challenge, and to explore practical strategies that enable safe AI adoption by influencing workforce behaviours.

METHODS AND ANALYSIS
This presentation draws on AI governance case studies across government, healthcare, and public sector organisations, combined with emerging lessons from real-world AI adoption patterns.

Rather than relying on traditional email surveys or time-intensive one-to-one interviews, our method involves the use of a digital human avatar interview tool designed to engage staff in a human-like conversation about their AI use.

The digital human avatar engages staff in structured, natural dialogue, efficiently revealing how employees use AI tools in practice and probing further where needed.

This approach reduces response burden, improves disclosure of informal or unapproved AI use, and provides richer qualitative insights into behavioural drivers.

Using a digital human avatar identifies:

First, a training and capability gap, where staff are unaware of approved tools or unclear on policy expectations.

Second, visibility of ongoing AI usage trends. Health care providers now require ongoing monitoring mechanisms to continuously assess AI usage as part of emerging governance obligations, rather than treating compliance as a one-off exercise.

The interview results are analysed against contemporary AI governance frameworks, including Australia’s AI governance standards and broader public sector risk management requirements.

OUTCOMES
Organisations that treat shadow AI as a behavioural and cultural challenge are better able to detect unauthorised use, reduce privacy and security risks, and support responsible innovation.

The use of conversational digital human avatar assessment tools provides a scalable way to surface real-world AI usage patterns, enabling targeted interventions such as training, policy reinforcement, and governance improvements.

When organisations act on these insights, they can improve policy adherence and gain clearer visibility of workforce AI use without compromising productivity or innovation.

FUTURE ACTIONS
Public health leaders should move beyond viewing Shadow AI as solely an IT issue and instead adopt a whole-of-organisation governance approach.

Future efforts should focus on building AI literacy, establishing clear behavioural expectations, embedding accountability within management structures, and implementing continuous monitoring mechanisms that reflect real-world AI usage.

The application of scalable conversational digital assessment tools can support this shift by providing ongoing visibility of workforce behaviour. By fostering a culture of responsible AI use, public health organisations can safely realise the benefits of AI while reducing the likelihood of privacy breaches, compliance failures, and reputational harm.

Background and Aim: The ongoing adoption of artificial intelligence in the Australian healthcare system raises concerns about the potential to perpetuate existing inequalities and disparities, particularly among underprivileged populations. Current studies in Australia have largely concentrated on system-level rather than community-level readiness. This study aims to examine how gaps in community readiness and digital inclusion may limit the effective adoption of AI-enabled healthcare in Australia.

Methods and Analysis: A preliminary structured literature review was conducted to identify and synthesise relevant studies from scientific databases (PubMed and Scopus) following PRISMA-ScR guideline. The analysis applies Van Dijk's digital divide framework, which includes mental, material, skills, and usage access, to examine multi-dimensional barriers affecting community engagement with AI-enabled healthcare in the Australian context.

Outcomes: Preliminary findings indicate that the digital divide in Australian healthcare operates across multiple dimensions. Mental access barriers include low trust in digital and AI-enabled systems, along with technology-related anxiety among certain populations, particularly culturally and linguistically diverse (CALD) and socioeconomically disadvantaged communities. Material access challenges persist where inadequate connections and a lack of suitable devices continue to be major obstacles to digital inclusion for many Australians living in rural and remote locations. Skills access gaps highlight the need for improved digital and AI literacy for individuals with limited digital skills. Usage access disparities are evident in the unequal adoption of telehealth and other digital health services across sociodemographic groups.

Conclusion and Future Actions: Despite the increasing AI adoption in healthcare, barriers across mental, material, skills, and usage dimensions persist at the community level. This paper contributes to global discourse around inclusive AI practice in the healthcare system. The findings can help inform policy and program stakeholders to design multilevel strategies in closing the gap in community readiness for the adoption of AI.