69238 research outputs found
Sort by
The Easy and Tough Aspects of School in New Zealand in the Middle Years
No full textStudent engagement and school attendance have been declining in New Zealand for some time, with some suggesting it has now reached ‘a crisis’ (Education Review Office, 2022). New Zealand schools also have high rates of bullying and challenging classroom behaviours compared to other OECD countries, along with declining literacy and numeracy levels (Education Review Office, 2024a; Jang-Jones & McGregor, 2019). Together, these negative trends suggest there is a need to try and understand why youth engagement in education is declining and what can be done about it. This study seeks to inform these discussions on how to improve student engagement and wellbeing by providing insights directly from rangatahi/youth about their experiences in school.
In this thesis I draw on data collected from 506 rangatahi in the Our Voices study who were approached via their participation in the demographically diverse Growing Up in New Zealand cohort. The rangatahi were asked questions about their school experiences via a co-designed app, with the questions analysed in my thesis being who has an easy and tough time at school and why, along with how school could better support those who have it tough. Taking a constructivist approach, I used reflexive thematic analysis to develop four main themes: ‘Seeing Through the Veil’, ‘It Depends’, ‘Fit in or Be Bullied’ and ‘School Structure Matters’. Together, these themes suggest that students in New Zealand have a deep understanding of the issues being faced by students at school, and they understand the variability and context dependent nature of these difficulties.
The students also suggested several ways in which they believe their experiences at school can be improved, including increasing the availability of counsellors, creating social clubs to help students make friends, implementing effective anti-bully measures, creating social skills or conflict management classes, improving student-teacher relationships and listening to the voices and feedback of students to collaboratively improve the school environment
Machine learning based clinical decision tool to predict acute kidney injury and survival in therapeutic hypothermia treated neonates
No full textTherapeutic hypothermia (TH) significantly reduces mortality and morbidities in neonates with Neonatal Encephalopathy (NE). NE may result in neonatal death and multisystem organ impairment, including acute kidney injury (AKI). Our study aimed to utilize machine learning (ML) methods to predict the outcome of TH-treated NE neonates developing AKI and death during TH. In this retrospective multinational study, 1149 TH-treated NE neonates and 801 controls were included. AKI was classified using KDIGO neonatal criteria based on serum creatinine measurements. The ML model incorporated gestational age, birth weight, postnatal age, and serum creatinine values. The algorithm used all these covariates to predict one of five outcomes: survival with/without AKI, mortality with/without AKI, and hospitalized non-NE controls. The XGBoost model achieved an AUC of 95% and an accuracy of 75.08% in predicting AKI and survival, surpassing other ML classifiers that demonstrated accuracy levels ranging from 54% to 65%. To our knowledge this is the first ML model trained on multicenter, multinational data specifically aimed at predicting neonates' AKI, death, and survival within the first three days. Our ML scoring systems' code and user interface are freely available ( https://github.com/NUBagciLab/Therapeutic-Hypothermia-Outcome-Classification , https://thprediction.streamlit.app/ ). This tool has potential to support neonatologists to personalize therapies, and to optimize pharmacotherapy for renally cleared drugs
Association of Antenatal Cytokine Concentrations with Neurodevelopmental Outcomes of the Offspring: A Scoping Review
No full textBackground: Maternal immune activation alters the intrauterine environment of the foetus, potentially affecting offspring neurodevelopment through the effects of maternal cytokines. We aimed to understand the scope of evidence available on the association of antenatal cytokines with offspring neurodevelopmental disorders (NDD) and outcomes (NDO).
Methods: A scoping review was done using Joanna Briggs Institute methodology and aligning with PRISMA-ScR guidelines. Six electronic databases and the grey literature were searched systematically. Human studies reporting antenatal cytokine concentrations and NDD/NDO were eligible.
Results: Of 4698 retrieved, 38 articles from 25 studies met inclusion criteria, with most being cohorts (18/25 (72%)) or case–control (5/25 (20%)) studies. Maternal samples were from serum (12/25 (48%)), plasma (8/25 (32%)) and amniotic fluid (3/25 (12%)). Most data were from second (21/38 (55%)) and/or third (22/38 (58%)) trimester samples. Most studies (29/38 (76.3%)) assayed maternal samples once in pregnancy, and the duration of sample storage ranged from 3 to 56 years. IL-6 was the commonest cytokine assessed (35/38 (92.1%)), followed by TNF-α (25/38 (65.8%)), IL-8 (22/38 (57.9%)) and IL-1β (20/38 (52.6%)). The majority reported NDO (26/38 (68.4%)), while autism spectrum disorder was the commonest NDD (11/38 (28.9%)) reported. Child age at assessment/diagnosis ranged from 6 days to 16 years. Most concluded that increased maternal IL-6 is associated with NDD and/or adverse NDO (19/35 (54.3%), but TNF-α, IL-1β, IL-4, IL-8, IL-10, and IL-17A had heterogeneous associations.
Conclusions: Maternal cytokines are associated with offspring NDD/NDO, but the findings are inconsistent due to heterogeneity in study designs and methodologies. Future research is needed to explore antenatal cytokines as an early biomarker for predicting offspring NDDs
Voyages of Wellbeing: Launching the Nesian Narratives Toolkit - A Pacific-Informed Wellbeing Resource for Early Childhood Education
No full textThis presentation introduces the Nesian Narratives Toolkit-a Pacific-led resource designed for kaiako
(educators) working with Pacific and non-Pacific children aged 4-5 in Early Childhood Education. Through
talanoa and talanga with over 50 Pacific parents, educators, health professionals, and community
members, the project explored culturally safe approaches to child wellbeing and sexuality education. The
collective insights revealed a common discomfort with explicit terminology, highlighting the need to reframe
these topics in developmentally appropriate and culturally affirming ways. In response, a co-design team
comprising two Pacific ECE educators, a community researcher, and an academic researcher created the
Nesian Narratives Toolkit: a strengths-based, health-educational resource organised into 13 themed
sections-an Introduction and "12 Voyages.”?Each Voyage presents planned learning experiences centered
on specific wellbeing topic, such as 'identity', 'brainworks', and 'body safety'. This presentation situates the
toolkit within place-based education, recognising sexuality learning as deeply relational and rooted in the
lived realities of children. It affirms the responsibility of educators and researchers to honour Indigenous
and Pacific ways of knowing, and to co-create education resources that foster connection, belonging, and
holistic wellbeing
How Children Share: Insights from the Sticker Task in 8-Year-Olds Growing Up in New Zealand
No full textSharing is a fundamental prosocial behaviour that fosters social cohesion, individual well-being, and interpersonal relationships. It is defined as a voluntary, intentional act of providing benefits to others at a personal cost without expectation of reciprocation. While individual differences in sharing are well-documented, the role of early socialisation experiences in shaping this behaviour remains underexplored. This study addresses this gap by identifying the strongest predictors of sharing in 8-year-old children using data from the Growing Up in New Zealand study, a large, nationally representative longitudinal cohort. Guided by Bronfenbrenner’s ecological model, we explored how individual, family, and broader social influences shape sharing behaviour at age 8 while also investigating whether early socialisation experiences at ages 2 and 4.5 predict later sharing. Sharing was assessed using the Sticker Task, where children could distribute stickers to an anonymous peer. Multivariable multinomial logistic regression analyses were conducted to identify key predictors across developmental stages. Findings indicate that sharing behaviour at age 8 is shaped by a complex interplay of individual, family, and social factors, with concurrent influences playing a stronger role than early socialisation experiences. Religious attendance was a robust predictor of fairness-based sharing, reinforcing moral socialisation theories. Additionally, children who rated themselves as highly prosocial were more likely to share, supporting Self-Perception Theory. Home-based early childhood education and birth order (being a subsequent child) also emerged as strong predictors, suggesting that early intimate socialisation contexts and sibling interactions promote fairness-oriented sharing. Unexpectedly, participation in sports and artistic extracurricular activities did not significantly predict sharing, suggesting that their role in enhancing prosocial behaviours may be more complex than previously assumed. In contrast, reflective and structured activities, such as parental book reading, arts and crafts, and television exposure, were positively associated with sharing, suggesting that engagement in perspective-taking contexts may reinforce fairness norms. This study contributes to prosociality research by integrating developmental frameworks to explore sharing behaviours. Findings suggest potential implications for parents, educators, and policymakers, highlighting the role of socialisation strategies in fostering sharing. Future research could further investigate the mechanisms underlying these associations and explore longitudinal pathways shaping prosocial behaviour beyond childhood
Automatic Bi-atrial segmentation and biomarker extraction from late gadolinium-enhanced MRI using deep learning
No full textAtrial fibrillation (AF) is associated with progressive structural remodeling of the atria, including chamber dilation, fibrosis, and variations in atrial wall thickness (AWT). Late gadolinium-enhanced (LGE) magnetic resonance imaging (MRI) has been used to quantify left atrium (LA) fibrosis for guiding adjunctive ablation beyond pulmonary vein isolation, though results have varied. A major limitation is the lack of a robust segmentation method for accurately assessing both atrial anatomy and fibrosis, coupled with the exclusion of the right atrium (RA) in the analysis. This study introduces biAtriaNet, a deep learning pipeline developed to automate segmentation of both LA and RA and to evaluate atrial fibrosis, AWT, and chamber diameter and volume from LGE-MRIs to support targeted AF ablation. biAtriaNet was trained and validated on 2D cine-MRIs from 4860 UK Biobank participants and 3D LGE-MRIs from 60 AF patients from the University of Utah, with independent testing on 11 3D LGE-MRIs at Waikato Hospital, New Zealand. The biAtriaNet consists of two CNNs based on a modified U-Net architecture with residual connections and batch normalization, optimized based on prior global benchmark study. This approach achieved accurate, consistent segmentation and biomarker extraction in UK Biobank and Utah datasets, validated against expert annotations. Additionally, biAtriaNet showed high transferability to independent datasets, achieving Dice scores of 91.1 % for LA and 88.6 % for RA. Chamber volume estimates closely matched ground truth values (LA: 89.8 ± 33.0 ml versus 91.1 ± 41.2 ml; RA: 70.8 ± 16.9 ml versus 72.3 ± 20.5 ml) with > 90 % accuracy in chamber measurements. AWT accuracies were 95.9 % for LA and 94.6 % for RA, while fibrosis estimates showed Kolmogorov-Smirnov correlations of 86.3 % (LA) and 90.6 % (RA) (p < 0.05). By enabling robust bi-atrial segmentation and biomarker extraction from LGE-MRIs, biAtriaNet has the potential to enhance patient-specific AF treatment strategies
Movement, Camouflage, and Background Complexity: Predator-Prey Interactions in Jumping Spiders
No full textThe interactions between predators and prey drive adaptations to improve the fitness of
individuals. Movement, camouflage, and background complexity are mechanisms that
influence these interactions and therefore, how adaptations evolve in both predators and prey.
However, understanding how these mechanisms interact to influence the success of predators
or prey has only recently been considered. This thesis examines how movement, background
matching camouflage, and complex backgrounds influence the detection and response by
predators and prey. To investigate prey, I used virtual targets moving at different speeds and
styles, and displaying various proportions of background matching. I assessed how these
factors influenced the detection and attack responses of the jumping spiders Maratus griseus
and Trite planiceps, and found that background complexity can significantly influence a prey’s
likelihood of detection. To investigate predators, I analysed how T. planiceps detected and
responded to virtual stimuli representing predators moving at different speeds across
backgrounds varying in complexity. I found that increasing background complexity may
reduce the detection of a predator, but incorporating movement as a component of a
backgrounds complexity may surprisingly increase detection. These findings highlight the
complexity of describing how movement, camouflage, and environmental backgrounds may
interact to influence the behaviours of predators and prey
Fostering Community Agency & Resilience to Prevent Sexual Violence Together
No full textIndigenous communities experience disproportionately high rates of sexual violence (UNFPA & CHIRAPAQ, 2018). The harms of colonisation across generations are often missing from conversations about our national identity and are not well understood in how they shape Māori experiences of sexual violence. There are particular vulnerabilities for Māori women (Fanslow, Robinson, Crengle, & Perese, 2007), young Māori women (Clark et al., 2016), and young takataapui (same sex-attracted) who are newly ‘out of the closet’ (Aspin, Reynolds, Lehavot, & Taiapa, 2009). Our study explored issues arising for rangatahi Māori who are becoming sexual beings through interviews with rangatahi Māori, kaimahi, and kaumātua. We were interested in how these issues were shaped by sexism, heterosexism and racism as well as possible solutions and positive representations of Māori sexuality through mātauranga Māori
Dynamic Resource Allocation for Network Slicing via GAT based Reinforcement Learning
No full textWith the rapid growth of emerging network transmission services, communication net- works face increasing challenges. 5G network slicing addresses these issues by enabling resource isolation and service customization through multiple virtual slices on a shared physical infrastructure. However, resource allocation and path selection in slice man- agement are NP-hard problems and must adapt to dynamic network states and service demands, making traditional static optimization methods inadequate. In recent years, reinforcement learning (RL) has gained widespread attention due to its effectiveness in high-dimensional decision problems. However, existing RL applications in slice re- source management still exhibit limitations in scheduling control, system modeling and state encoding when applying reinforcement learning to slice resource management. Current advanced methods leverage graph neural networks and continuous output reinforcement learning techniques.
To address these challenges, this thesis proposes a novel network slicing resource allo- cation framework. The method incorporates a state encoding module based on Graph Attention Networks (GAT), which effectively captures dynamic network topologies and link attributes. It employs the Deep Deterministic Policy Gradient (DDPG) al- gorithm to generate continuous action outputs for both bandwidth allocation factors and path weights, while integrating the * algorithm to provide heuristic guidance for path selection. This framework features a clearer task division and outperforms conventional graph neural networks. Experimental results show that the proposed framework achieves better performance than baseline methods in terms of resource fairness and latency reduction. Delay improvement is about 16.7%. This intelligent scheduling framework improves service differentiation by prioritizing traffic based on real-time demands, aids researchers in testing AI-based networking methods, and en- hances user experience through better reliability and responsiveness
Privacy in Smart Health Monitoring: A Systematic Review and Research Directions
No full textPrivacy concerns related to surveillance technologies are a primary deterrent for consumers hesitant to share their health data with service providers in smart health monitoring systems (SHMSs). These concerns can impede the adoption and operational success of SHMSs, leading to dissatisfaction among both consumers and service providers. Despite the significance of privacy, existing literature on SHMSs tends to offer a somewhat fragmented exploration of this concept due to the complex nature of surveillance and the involvement of multiple stakeholders. To address this gap, this study develops a contextual framework based on a systematic review of 49 peer-reviewed articles, offering valuable insights for scholars seeking to understand the multifaceted privacy concerns in SHMS contexts. The findings emphasize the importance of integrating theoretical perspectives that better capture the intricate dynamics of smart health environments, helping healthcare providers and policymakers identify and address potential privacy issues when developing and implementing surveillance systems for personal health information. Additionally, the study highlights existing knowledge gaps and proposes six research avenues to achieve a deeper understanding of privacy in SHMSs