Meena Jha

This study investigates the integration of Indigenous perspectives within information and communication technology (ICT) curricula across Australian universities, addressing four research questions relating to the extent of integration, approaches used, influencing factors and characteristics of effective practice. Semi-structured interviews were conducted with eight ICT educators and data were analysed using reflexive thematic analysis. Nine themes were identified. Findings indicate that integration remains limited and inconsistent, often driven by individual initiative rather than systematic institutional approaches. While universities demonstrate policy commitments, these are not consistently enacted in curriculum design. Educators employ strategies such as contextualised assignments, interdisciplinary offerings and emerging Indigenous pedagogical frameworks. Key barriers include limited discipline-specific resources, structural curriculum constraints and lack of targeted support. Enablers include collaboration with Indigenous experts and culturally relevant teaching approaches that enhance student engagement. The study offers insights to inform curriculum development, institutional policy and future research in ICT education.

Global interest in artificial intelligence- (AI) supported assessment is growing; however, few studies have examined how institutional responses align with regulatory standards, particularly in information and communications technology (ICT) education. This study conducted a national-level document analysis of publicly available resources from 39 Australian universities offering ICT programmes, exploring how institutions are responding to generative AI integration in assessment practices against the Tertiary Education Quality and Standards Agency’s two core principles: preparing students to engage ethically in an AI-driven world and ensuring reliable and trustworthy assessments of student competencies. As ICT-specific documentation was largely absent from the public domain, findings were drawn predominantly from generic university-wide policies, with ICT serving as a case context where discipline-specific guidance is notably lacking. Findings revealed encouraging progress in institutional positioning, with publicly available documents indicating increased policy articulation, referenced training initiatives and stated commitments to AI literacy. However, the analysis also exposed inconsistencies in policy coherence, limited ICT-specific guidance and substantial variation in how GenAI is addressed across institutional documents. The paper recommends stronger integration of AI ethics, clearer communication of permitted AI uses and targeted professional development for ICT staff, contributing to evolving discourse on responsible AI use and equitable, future-ready assessment design. Implications for practice or policy: ICT course leaders should develop discipline-specific GenAI guidance for technical assessments, largely absent across Australian universities. Institutional leaders should invest in operational governance structures, documented in only 7 of 39 universities. Sector bodies should develop discipline-specific extensions to existing GenAI frameworks, as institution-level principles have not produced targeted ICT guidance. Assessors should prioritise process-based and authentic ICT assessment designs, referenced by only 21 universities.

While deep learning has emerged as a powerful tool for automated medical image analysis, existing Convolutional Neural Networks (CNNs) often struggle with limited datasets and the "vanishing gradient" problem when scaled to deeper architectures. In this study, we propose a robust, high-precision Deep Transfer Learning framework designed specifically for the multiclass staging of Alzheimer’s disease using structural Magnetic Resonance Imaging (MRI). We utilized a data set of T1-weighted magnetic resonance imaging scans from the Open Access Series of Imaging Studies (OASIS), rigorously balancing the distribution across four classes: Non-Demented, Very Mild Demented, Mild Demented, and Moderate Demented. To overcome the limitations of training deep networks on small medical datasets, we employed a ResNet50 architecture pre-trained on ImageNet. This extensive backbone leverages residual skip connections, allowing the network to learn identity mappings (F(x)+x) and effectively propagate gradients through 50 layers without degradation. We further customized the architecture by replacing the final classification head with a specialized block containing Global Average Pooling, discrete dropout layers (p = 0.5), and dense layers to prevent overfitting and enhance feature extraction capabilities. Our experimental results demonstrate that the proposed model achieves a state-of-the-art overall classification accuracy of 94.56% and a macro-averaged F1-score of 94% on a held-out test set. Crucially, the model exhibits high sensitivity in distinguishing the ’Very Mild Demented’ class from ’Non-Demented’ controls, a task where traditional shallow networks and human radiologists frequently underperform. This superior performance validates the efficacy of deep residual learning in capturing non-linear, hierarchical patterns of cerebral atrophy. By automating the extraction of these complex biomarkers, our proposed model offers a reliable, scalable, and objective second opinion for clinicians, potentially varying the trajectory of patient care through earlier and more accurate diagnosis. Experimental results demonstrate that the proposed framework achieves an overall multi-class accuracy of 94.56%, yielding a performance improvement of 1.38% to 5.16% over comparable state-of-the-art multi-class baseline models. Significantly, the model attains an exceptional specificity of 0.98 and high sensitivity for the ’Very Mild Demented’ stage, addressing a critical clinical need for reliable, early-stage Alzheimer’s detection when therapeutic interventions are most effective.

In an era where data-driven decision-making is paramount, “Decoding Data: A Complete Guide to Business Intelligence” serves as an essential Open Educational Resource (OER) tailored for both students and professionals seeking to develop a deep understanding of Business Intelligence (BI) strategies. This resource aims to empower learners with the knowledge and skills necessary to harness the full potential of BI in the modern business landscape. This comprehensive resource covers a wide spectrum of BI topics, including data to knowledge continuum, systems for decision support, the architecture of BI, business analytics, reporting, and emerging technologies. An innovative pedagogical approach combines theoretical concepts with practical skills for effective data utilisation by creating reports, visualisations, and dashboards, offering readers a holistic understanding of BI concepts. With hands-on exercises and real-world case studies, “Decoding Data: A Complete Guide to Business Intelligence” goes beyond traditional learning methods, providing interactive tools that engage and challenge the reader. The content is organised as a roadmap, guiding readers through the complexities of BI implementation, starting from foundational concepts and progressing seamlessly toward advanced data analytics. This journey equips users with theoretical groundwork and practical skills directly applicable across diverse business scenarios. The book’s focus on key concepts and theories ensures users a sophisticated comprehension of BI, enhancing their analytical skills and decision-making capabilities. More than a textbook, “Decoding Data: A Complete Guide to Business Intelligence” is a pathway to enhancing analytical skills, refining decision-making abilities, and cultivating a profound appreciation for BI’s pivotal role in driving organisational success. This resource promises not just knowledge acquisition but tangible skills development. Embark on a transformative journey with “Decoding Data: A Complete Guide to Business Intelligence”, where theory meets practice, and education evolves into actionable insights.

The advent of generative Artificial Intelligence (Gen-AI) has transformed the way higher education is delivered. Gen-AI is rapidly changing how students engage with learning resources, with academics adopting these tools in the classroom to provide insightful lessons and scalable solutions that enhance educational outcomes across disciplines. We use the Technological Pedagogical Content Knowledge (TPACK) framework to discuss the types of knowledge teachers require to effectively use GenAI tools in their teaching. This paper reports on five case studies from an Australian higher education institution examining the integration of Gen-AI across various academic disciplines in higher education, demonstrating its impact on teaching methodologies and creating innovative educational experiences. Throughout these diverse case studies, we show how Gen-AI tools can be used to enhance academic skills, foster creativity, and address pedagogical challenges. Each case study reveals the complexities and potentials of Gen-AI in educational settings, including the development of critical thinking skills and understanding Gen-AI’s limitations. We describe how Technological Knowledge, Pedagogical Knowledge and Content Knowledge of teachers has impacted the use of Gen-AI in the classrooms.

A regional university in Australia is dedicated to developing 21st-century graduates as global citizens. The curriculum is designed to equip students with the knowledge and skills necessary for success in a globalised world. Assessing whether students are truly becoming ‘global citizens’ involves evaluating not just academic performance, but also intercultural competencies, global awareness, and adaptability. Immediate, tangible outcomes like job placements can take precedence over long-term goals like global readiness. There are many factors that hinders a comprehensive review of whether the curriculum is effectively preparing students to become global citizens. There has not been any systematic understanding of the research status of the internationalization of the curriculum (IoC). This paper reports on the experiences of academics and students regarding the internationalisation of the curriculum within the Faculty of Arts and Education (FoAE). A mixed-method approach was adopted over two years, involving surveys and one-on-one interviews with academics and students. Thematic analysis of the responses revealed both strengths and areas for improvement for integrating IoC in higher education institutions. While academics and students value an internationalised curriculum, it is inconsistently implemented across the curriculum and there is a need for universities to establish a clear framework or benchmark for evaluating the internationalisation of their curriculum to assess progress systematically. The study's findings offer valuable insights for enhancing the internationalisation of the curriculum and improving the effectiveness of academic staff in delivering globally relevant content. This could lead to a stronger institutional reputation, attract a more diverse and international student cohort, and has capability to positively impact the university's financial sustainability by aligning the curriculum with global education trends.

The identification of malaria infection using microscope images of blood smears is considered as a ‘gold standard’. The diagnosis of malaria needs expert microscopists which are scarce in remote areas where malaria is endemic. Therefore, it is desirable to automate the repetitive task of pathogen detection in the blood samples received as microscope images. This study provides an easy to use and deploy method for implementing a malaria pathogen detection software- the Intelligent Suite. The Intelligent Suite features a graphical user interface (GUI) implemented using ‘cvui’ library to interact with the OpenVINO’s inference engine for model optimisation and deployment across several inference devices. The intelligent Suite uses a custom YOLO-mp-3l model trained on Darknet framework for detection of malaria pathogen in thick smear microscope images. Moreover, the Intelligent Suite provides user interface for inference device/mode selection, alter model parameters, and generate detection reports along with the model performance metrics. The Intelligent Suite was executed on a CPU computer with model inference running on a plug-and-play Neural Compute Stick (NCS2) and performance reported.

Australia, like many other countries around the world, is undergoing a transition toward net zero emissions. It requires changes and development in many sectors, which not only bring benefits but also challenges. The rapid growth in renewable energy sources (RESs) is necessary to decarbonise electricity generation but negatively affects grid stability. Residential buildings also contribute to this issue through specific load profiles and the high penetration of rooftop photovoltaic (PV) installations. Maintaining grid balance will be crucial for further emissions reductions. One of the potential solutions can be the replacement of conventional heating and cooling systems in houses with solutions capable of storing energy and shifting the electrical load. As presented in this paper, heat pumps and hydronic systems can significantly improve the electrical load of a typical South Australian household when they are controlled by algorithms reacting to the current grid conditions and household-generated electricity compared to conventional solutions. TRNSYS 18 simulations of air source and ground source heat pump systems with smart control based on measured electricity consumption and domestic hot water usage data showed the possibility of total energy consumption reduction, shifting the load from peak periods towards periods of excessive RES generation and increasing self-consumption of rooftop PV electricity. These improvements reduce the amount of emissions generated by such a household and allow for further development of other sectors.