Welcome to Research Commons

Research Commons is the University of Waikato's open access research repository, housing research publications and theses produced by the University's staff and students.

Communities in Research Commons

Select a community to browse its collections.

Now showing 1 - 2 of 2

Recent Submissions

Architecting a co-designed clinical decision support system for diabetes management
(ACM, 2026) Wang, William Yu Chung; Jiang, Philip Hong Wei; Shafiu, Ibrahim; Paul, Ryan; Keenan, Rawiri; Madurapperumage, Anuradha; Liu, Chen
Diabetes mellitus, particularly Type 2 diabetes (T2D), imposes a substantial and inequitable burden on New Zealand, with domestic communities disproportionately affected. Early prediction of diabetes complications using machine learning (ML) has shown strong potential, yet most models remain disconnected from real-world clinical workflows and health information systems. This study outlines a co-design and design science research (DSR) programme to develop a machine learning–enabled clinical decision support (CDS) framework for predictive diabetes management in the Waikato region. Working with Health New Zealand (HNZ) Waikato, clinicians, and health informaticians, the research proceeds through three stages: situation awareness of current diabetes management processes, co-design of a data-driven CDS architecture, and evaluation via predictive performance metrics and process simulation. Expected contributions include an empirically grounded CDS framework tailored to local clinical and cultural contexts, improved predictive capabilities for T2D complications and resource planning, and a scalable pathway toward AI-enabled diabetes services that support health advancement and health equity.
Identifying and categorising metaphors in New Zealand breast cancer discourse: Methods and challenges
(2026) Malik, Sara; Calude, Andreea S.; Ulatowski, Joseph W.
Metaphor is one of the common features in health communication, particularly in cancer discourse, as analysed by Semino et al (2018). While studying metaphors offers valuable insights, the process of identifying and categorizing them is challenging (Glynn and Biryukova, 2022). In this paper, we outline methods of metaphor identification and categorization in the YouTube breast cancer discourse of New Zealand English and also discuss the challenges involved. We collected a corpus of forty-six personal narratives from New Zealand, posted between 2011 and 2023 on YouTube by breast cancer organizations. The data was transcribed using the Whisper AI tool and curated to include only patients’ discourse. This presentation discusses: (a) methods of metaphor identification, (b) methods of metaphor categorization, and (c) challenge in metaphor categorization and its solution. For identification, we used Conceptual Metaphor Theory (Lakoff and Johnson, 1980), the Metaphor Identification Procedure (Pragglejaz Group, 2007), and the New Zealand Oxford Dictionary to determine the basic meanings of words. For metaphor categorization, we used two approaches: (a) source domain labels, for instance, violence (terms representing any form of physical harm are categorized under the ‘violence’ category, such as ‘fight’, ‘hit’, ‘aggressive’) and (b) recurrent categories identified in existing literature. One challenge was that some metaphors were borderline and could not fit neatly into a single category. To address this, we classified them as extensions of an existing category. For example, the term ‘fight’ clearly denotes physical harm and fits within the violence category, whereas ‘death sentence’ reflects punishment rather than direct violence. Therefore, it is considered an extension of the violence category, named as punishment metaphor. This study offers a systematic approach to metaphor identification and categorization, which can often be a challenging task.
You're killing me
(The University of Waikato, 2026) Blood, Andrew; Slaughter, Tracey; Chidgey, Catherine
If we build superintelligence, the joke is on us. By the 2050s, when most human skills have been replicated and automated by artificial intelligence, humanity has one last trick up its sleeve. Humour. Because AI doesn’t have a funny bone. Yes, it might be super intelligent, but it cannot fathom how or why jokes work. It just doesn’t get it. Worse, jokes expose bugs in its operating system and any AGI trying to understand them gets tied up in recursive knots. Comedy is an existential threat. Likewise, the authoritarian regimes that control the algorithms behind AI also feel threatened by comedians because comedy, amongst many other things, often contains deep human truths. And that’s something that scares them. Therefore, to the authorities and artificial intelligence alike, the comedian is the most dangerous person in the room. You’re Killing Me, is a speculative fiction novel. A twisted retelling of the Pandora myth, where all the techno plagues released upon humankind refuse to go back into their black boxes, and our last remaining hope comes courtesy of those who are prepared to stand up. Page by page, we are drawn into a world that is ever more complex and disorienting, where omnipresent AGI surrounds us, and synthetic human companions called Pandoras live alongside us. We work with them, fall in love and make love to them, and even reproduce with them. Which also means we commit adultery with them, fight with them, divorce and despise them, and are openly species-ist towards them. In fact, everything we do to humans we do to them, so their arrival, instead of making us happier, only multiplies the ways in which we are unhappy. To make matters worse, synthetic human population growth outstrips that of humans, and the colossal data centres that power them push our planet’s resources to the limit. ‘Machines drink first. Humans drink second,’ is the law. The only sane response is to laugh. And thankfully, comedy is a weapon. Humour has always been a survival mechanism, and now it’s our only chance of survival. The battle between humanity and AI is set. The question is, who will have the last laugh? The future’s dark. But at least it has a funny side.
From user feedback to iterative decision-making: A low-burden multi-channel feedback framework for mobile health applications targeting non-communicable diseases
(ACM, 2026) He, Yuyu; Wang, William Yu Chung; Zahid, Arnob
Mobile health applications (mHealth Apps) can support the long-term management of non-communicable diseases (NCDs). However, their practical value is commonly limited by low user dropout rates and declining engagement over time. To identify meaningful directions for further improvement, a more comprehensive user evaluation is needed. Existing evaluation methods generally rely on questionnaires and interviews, which can provide valuable quantitative and qualitative evidence. At the same time, studies on user-generated content (UGC) show that free-text feedback can also reveal specific directions for product iteration. Yet these sources of feedback are often collected and analysed separately. Further exploration is required on how to integrate multiple feedback sources while minimising the burden on users with limited time and energy, thereby generating actionable evidence to support subsequent iterations. Therefore, this study proposes a low-burden, multi-channel user feedback framework for NCDs mHealth Apps. It integrates three types of feedback—questionnaires, interviews, and UGC—to obtain more comprehensive user feedback evidence and provide a basis for subsequent iterations. The framework comprises three stages: feedback collection and problem identification, cross-source evidence collation, and iteration-oriented outcome delivery. It aims to gather more comprehensive, timely user feedback, facilitating product iterations that better align with user needs and laying the groundwork for future validation in real-world settings.
The impact of data aggregation: Advocating for individualized analysis in wearable sensor research
(ACM, 2025) König, Jemma; Penaredondo, Jascha; Lim, Nick Jin Sean; Hinze, Annika; Bowen, Judy; Viller, Stephen; Paay, Jeni; Fredericks, Joel; Turner, Jane; Vickery, Nicole; Wadley, Greg; Muñoz, Diego; Capel, Tara; Atiq, Arzoo; Davis, Hilary; Boden, Marie; Harman, Joel; Ploderer, Bernd
The use of wearable sensors to record physiological data is becoming increasingly common, both in research and in daily life. User studies collect data using wrist bands, chest straps, and headbands to measure heart rate, skin conductance, and brain activity, to name a few. These readings can then be used to classify a range of physical and cognitive functions, such as cognitive and physical workload, cognitive and physical fatigue, stress, and attention. However, while physiological data is highly individualized, many researchers use machine learning classification methods that combine the participants' data into one dataset. In this paper, we demonstrate the negative impact this has on results and show that the individualized nature of physiological data requires individualized analysis and classification.