The Research Trust of Victoria University of Wellington Smart Ideas funded projects

Radar systems, quantum computing, and other high-frequency applications all rely on devices called microwave circulators, which send microwave signals preferentially in one direction only. But microwave circulators are bulky. Quantum computers cannot be scaled up because of the physical limitation on the number of circulators that can be packed into the cold space used to perform quantum computations.  This is a fundamental roadblock for large-scale quantum computing. We aim to overcome these challenges by designing miniaturised circulators that can be fabricated directly on semiconductor chips. We will use advanced materials research and cutting-edge fabrication techniques to design and build new devices that overcome the problems of size and the stray magnetic fields that beset current circulator designs. This is highly novel, challenging research. To succeed, we will work with leading microwave device researchers in Australia and microwave measurement experts at the University of Otago. These partnerships will allow us to rapidly test the materials and the device designs required for this application and develop proof-of-concept devices that can be optimised to compete with current technologies. One of the outcomes of the programme is to develop these on-chip circulators further to bridge the gap between the lab and genuine practical commercial uptake. To this end, we are being advised by international quantum computer hardware companies on the specific metrics for the devices so that they can be incorporated into designs for scaled-up quantum computers in future. If successful, our research will put New Zealand into a position to participate in the international supply chain for quantum computing. The quantum computing market is estimated to exceed US$12 billion over the next decade and will continue to grow strongly into the future.

The National Seismic Hazard Model (NSHM) is crucial for reducing the impact of large earthquakes in Aotearoa-New Zealand. The latest NSHM highlights the risks from extreme shaking, but significant uncertainties complicate decision-making for engineers, planners, and insurers. To improve accuracy, NSHMs need long-term records of earthquake shaking, which modern sensors can't provide. Our innovative solution uses lake sediments as natural "seismometers" to create earthquake records spanning thousands of years. Strong earthquakes cause lake-margin sediments to collapse, leaving layers in the lakebed that serve as historical fingerprints of seismic activity. By analysing these layers, we will quantify past shaking intensities and refine the NSHM. The project focuses on two lakes, Rotoroa and Tūtira, which record earthquakes from the Alpine Fault and Hikurangi Subduction Zone. We will use advanced tools like fibre-optic seismology to measure the earthquake intensities needed to disrupt lake-margin sediments, and sedimentary DNA to identify which landforms contributed to sediment layers during past earthquakes. Our method will be validated with known earthquakes recorded in the lakes, then applied to long-term sediment records to refine absolute hazard values and reduce uncertainties in the NSHM. Our results will be rapidly implemented through our team's representation on the NSHM advisory board and through existing relationships with key stakeholders, including local and central government agencies (e.g., NZ Transport Agency, and regional councils). This groundbreaking approach aims to improve global earthquake modelling and enhance seismic risk management in Aotearoa-New Zealand. It will provide economic and social benefits by enhancing earthquake preparedness in Aotearoa-New Zealand. Through an Iwi Māori-focused education programme, we aim to inspire tamariki and rangatahi to explore careers in geoscience, earthquake engineering, and emergency management, fostering resilience and knowledge in future generations.

Many aspects of New Zealand's society and economy are highly exposed to volcanic eruptions. If we are able to accurately forecast future eruptions, we would be able to mitigate much of this risk. Volcanoes are monitored using multiple different measuring techniques. These all attempt to detect changes in the volcano which may indicate an impending eruption. Volcano scientists use all of these data streams simultaneously to try and forecast the probability of an eruption. This is highly challenging and subtle changes in the data can often be missed. Our project will develop new machine-learning tools to jointly analyse all of the available data at Aotearoa New Zealand's volcanoes. These new tools will be able to detect anomalies in the volcano monitoring data and use this to accurately forecast the probability of an eruption at any of our active volcanoes. Contact email address: finnigan.illsleykemp@vuw.ac.nz

Where will climate change accelerate or slow river mobility in Aotearoa? We know that many of Aotearoa’s large rivers gradually shift across their floodplains over time, eroding their banks a little more with each major storm that passes through their headwaters. We also know that river flows and the size of floods are likely to increase in many of our awa as climate changes in this century. But we do not yet know how these projected climate changes might alter the pace of river mobility in Aotearoa, nor whether current approaches to river management will be suitable in the face of climate change. Our project sets out to answer these questions. We will measure current mobility rates in our large rivers using satellite data stretching back to the late 1980s. Then, we’ll use machine learning to test how much these river mobility rates are due to river flows and flood frequency, as well as other important controls like stopbanks, riverbank vegetation and the sediment carried by each river. Finally, we’ll combine these findings with NIWA’s projections for future river flows to assess where in Aotearoa we might expect river mobility to accelerate or slow down as climate changes. Our results have important implications for ensuring we have safe setback zones along rivers and leave rivers enough space to migrate and maintain their natural ecosystem without endangering humans through flooding or riverbank erosion.

Cancer therapy increasingly depends on the use of drugs for specific anticancer targets. Cancer resistance to these therapies is becoming common. New approaches that have the potential to attack resistant cancers and prevent development of drug-resistance are desperately needed.

DNPH1, a newly discovered anticancer target in breast and ovarian cancers hold great promise to provide new avenues of anticancer therapy in combatting drug resistance. This precision medicine approach can complement and strengthen traditional therapy by re-sensitising cancers to the therapy. No drug is available for this target.

This research project plans to develop drug candidates for the new drug target DNPH1. Our approach of transition state theory to discover potent and specific drugs has the potential for the rapid development of new anti-cancer agents for chemo-resistant breast and ovarian cancers.

The reintroduction of kākā to Wellington City has gifted a taonga to Te Whanganui-A-Tara, but also created new conservation challenges. As kākā spread beyond Zealandia Sanctuary and into urban areas, our ability to identify and monitor individuals is hampered by a reliance on traditional approaches such as bird banding. Very few kākā are currently banded in Wellington, and so we have very little information on the movement and survival of birds in the city. This lack of individual-based knowledge impacts our ability to mitigate new threats to urban kākā, including diseases, unintentional poisoning, and conflict with humans.

We seek to create a new AI-based approach for identifying individual birds, in partnership with tangata whenua (Taranaki Whānui) and conservation organisations. We will develop an AI tool that will be capable of identifying many more individual kākā in the Wellington population than currently possible, adding valuable knowledge on the behaviour and movement of birds in the city. In doing so, we will expand mātauranga and our ability to effectively mitigate new threats to this taonga.

Working out how many animals are in a population is important but challenging, particularly for our remote, hard-to-see taonga species in Aotearoa. Since many of our bird species have loud calls, acoustic monitoring is a standard method to keep tabs on them. However, successful wildlife management needs more than just knowing if they are present: we need to be able to estimate the size of the population. And that means differentiating between one bird calling many times, or lots calling once each.

Since many species appear to be able to recognise each other from their calls (pairs might duet together, and in some species newcomers to an area can expect an aggressive visit from the locals if they dare to call) we ask if we can train machines to do it too. If so, we can incorporate that information into statistical tools to track changes in population size over time for our taonga species and help them thrive.

In this project we will focus on kiwi, an iconic species that many iwi and community groups across Aotearoa work very hard to help. We will develop knowledge, methods, and tools that automatically identify individual kiwi from their calls and use that to identify the number of birds present. We will test our methods by generating fake kiwi calls of new individuals and playing them to the inhabitants of the forest to see how they respond, and by comparing the estimates of population size we obtain with those from more intrusive methods such as dog surveys. We will also ask if there is a link between the call and the genetic health of the birds.

Congestive heart failure is a chronic and life-threatening disorder with enormous clinical, social and economic impact that affects millions of people worldwide. It is a leading cause for hospital admissions and many patients die within five years of diagnosis. To date, HF has no cure – the disease can only be managed through pharmacological intervention which is mostly aimed at treating symptoms. A primary factor in the progression of the disease is the overactivation of the cardiac mineralocorticoid receptor. Therefore, mineralocorticoid receptor antagonists (MRAs) would be ideal drugs, if it weren’t for their deleterious side-effects that cause a detrimental rise in potassium levels in the blood and an acute deterioration of renal function. These side effects are a result of the systemic, non-selective actions of current MRAs that also disrupt the accurate function of the mineralocorticoid receptor in the kidney. Thus, many patients are precluded from benefiting from this therapy option.
 
To address this issue, scientists at the Ferrier Research Institute and the Baker Heart and Diabetes Institute in Melbourne, Australia, have now teamed up to create a new MRA drug with a selective mode of action that will protect the heart but also spare the kidneys. Without previous limitations attached, this drug will be safe and therefore well-positioned to cater to the needs of a wide range of heart failure patients. The drug will be patent-protected and available for development by a New Zealand biotechnology company, in that way growing New Zealand’s knowledge-intensive biotech industry, creating high-skilled employment, and prompting high-value product manufacture by existing New Zealand companies.

We will develop an Artificial Intelligence-based model that can predict sea level rise over the next several thousand years. Our approach is based on Evolutionary Learning, and unlike other AI approaches, can be physically explained for the processes that lead to the prediction. Our team consists of experts in Evolutionary Learning, AI, interpretable AI and in Earth systems processes, specifically climate change processes. Current models that calculate sea level rise from melting ice sheets must carefully balance the desired accuracy based on modelling the multitude of complex processes that influence the melt process with computational efficiency. Due to the computational costs inherent in traditional models, long-term accurate predictions over several millennia are missing.  Our model will overcome these shortcomings.

We will deliver our new algorithms in a visualisation tool that allows in an interactive interface to correct simulation errors and improve the model.

Plants and microbes produce natural products to protect themselves against other organisms. These natural products have found many uses in modern society, notably as antibiotics for treatment of infectious diseases. New technologies are required to accelerate natural product discovery and development of natural products to provide solutions for modern problems. For example, many diseases are becoming resistant to existing antibiotics, limiting our ability to treat them.

Fungi have provided many of the natural products that are used as antibiotics or other medicines today. Studies have shown that fungi contain enormous untapped reserves of undiscovered natural products. However, most of these products are not produced under laboratory conditions, making them difficult to obtain. This program will develop technologies that allow us to produce any fungal natural product within the laboratory, accelerating the rate at which these molecules can be discovered and bought to market.

Having successfully demonstrated proof-of-concept for our technology during the primary contract period, this extension will focus on refining and commercialising this technology to generate real economic impact for New Zealand.

Plants and microbes produce natural products to protect themselves against other organisms. These natural products have found many uses in modern society, notably as antibiotics for treatment of infectious diseases. New technologies are required to accelerate natural product discovery and development of natural products to provide solutions for modern problems. For example, many diseases are becoming resistant to existing antibiotics, limiting our ability to treat them.

Fungi have provided many of the natural products that are used as antibiotics or other medicines today. Studies have shown that fungi contain enormous untapped reserves of undiscovered natural products. However, most of these products are not produced under laboratory conditions, making them difficult to obtain. This program will develop technologies that allow us to produce any fungal natural product within the laboratory, accelerating the rate at which these molecules can be discovered and bought to market.

This project brings together artificial intelligence (AI) and drug discovery to develop novel and selective kinase inhibitors. Kinases are enzymes that play crucial roles in cellular signalling pathways, and kinase inhibitors have shown strong potential in the treatment of diseases such as cancer and metabolic disorders. However, achieving target selectivity remains a major challenge in developing new therapeutic agents, as there are more than 500 closely related kinases encoded in the human genome.

The project aims to develop an AI framework that designs kinase inhibitors with high target selectivity and prioritises those most likely to be effective. By integrating experimentally determined 3D structures of kinase binding pockets into advanced AI models, the framework can analyse the precise structural and physicochemical features that govern inhibitor selectivity and potency. This structure-guided approach enables the design of compounds with the inherent characteristics of highly selective inhibitors.

This transformative research has the potential to revolutionise pharmaceutical discovery, accelerating the development of safer and more effective therapies. In addition, it will contribute to the growth of high-value biotechnology and pharmaceutical industries in New Zealand, strengthening connections to international research networks and markets. The AI-driven approach also supports a low-emission, sustainable research environment by reducing reliance on energy-intensive experiments and minimising chemical waste.

This project brings together artificial intelligence and drug discovery to develop novel and selective kinase inhibitors. Kinases are enzymes that play a crucial role in signalling pathways within cells, and kinase inhibitors have shown potential for treating diseases such as cancer and metabolic disorders. However, the issue of target selectivity has been a significant challenge in the development of novel therapeutic agents.

This project aims to develop an artificial intelligence framework that will design kinase inhibitors with high target selectivity and prioritise those with the potential to be effective. By exploring molecular features that engender selectivity in kinase inhibition, the framework will be able to produce compounds that have the inherent characteristics of target-selective inhibitors.

This transformative research has the potential to revolutionise pharmaceutical discovery, opening up new possibilities for drug development. Additionally, the project will contribute to the development of high-value industries in New Zealand, such as the biotech and pharmaceutical sectors, thereby diversifying the economy and linking it to international expertise and markets.

The proposed approach to drug discovery is also environmentally friendly, contributing to the transition to a low-emission economy. By reducing the use of energy-intensive experiments and environmentally harmful chemicals, the method enhances the efficiency of drug discovery while contributing to environmental sustainability.

Artificial Intelligence (AI) is transforming the personal and professional lives of everyone in NZ.

Most of the important advances in AI use deep learning (DL). DL systems are now considered indispensable in many industry sectors. But in some sectors, lack of transparency limits uptake. The way a DL system makes a decision is not transparent to the user.

The active research area of explainable AI (XAI) attempts to provide this transparency. Many XAI approaches attempt to “look inside” the black box. These approaches cannot leverage the power of DL.

We have prototyped a different XAI concept. We use repeated querying of generative models to measure the information that most impacts the decisions. This project turns the concept into practical applications. In particular, the research

  (a) adapts the concept to the generative models applicable in particular use-cases, and

  (b) demonstrates the reliability of the outputs and the quality of the explanations provided.

A target application of our XAI system is the detection of mis/disinformation in NZ political social media. The research includes a detailed evaluation of the system at national scale by comparing it with leading baselines and human coders. The election is a particularly high-impact use-case for NZ society.

The research also addresses two specific problems in the area of disinformation detection. The first is the need to check factuality, rather than simply correlated traits, such as tone or source. The second is the growing incidence of “half-truths” in social media. These are true facts designed to lead to false conclusions. Half-truths are often more deceptive than false information. They have proven to be difficult to detect, for both humans and machines; we will target addressing this difficulty.

Artificial Intelligence (AI) is transforming the personal and professional lives of everyone in NZ.

Most of the important advances in AI use a technique called deep learning (DL). DL systems improve productivity and safety in manufacturing, healthcare, agriculture, finance, government, and research.

The main impediment to the uptake of AI is lack of transparency. DL systems are notorious for being 'black boxes', i.e., the way that a decision is made is not transparent to the user. Users are right to be wary of decisions that are not explained. Governments have legislated the right to “meaningful information about the logic involved” for people affected by an AI decision but cannot provide it.

The active research area of explainable AI (XAI) attempts to provide this transparency. Many XAI approaches attempt to “look inside” the black box. These approaches cannot leverage the power of DL.

We have prototyped a different concept which uses repeated querying of generative models to reveal and measure the information that most impacts the decisions. This project aims to develop the concept into practical applications. The research will:

• improve its speed
• adapt it to the generative models applicable in particular use-cases
• prove its competitiveness and effectiveness in terms of the reliability of the outputs, and the quality of the explanations provided.

The research will focus on a concrete use-case, namely disinformation in social media, in partnership with the New Zealand Social Media Study (NZSMS). The NZSMS promotes an informed public by publishing findings of disinformation in political party social media. NZSMS currently uses manual text analysis and use of this technique will ensure more efficient, rapid, and explainable analysis.

In 2019, the U.S. Federal Communications Commission estimated that improving the average ambulance response time by sixty seconds would save 10,120 lives per year, each valued at US$9.1 million. We infer that such an improvement projected to New Zealand could save up to 150 lives per year, each valued at $4.46 million by the Ministry of Transport.

The problem faced by ambulance services changes daily: they do not know how many emergencies of which urgency will occur when or where. New Zealand services act with a limited set of resources and struggle to meet the response time targets set by Government. While an increase in infrastructure funding would improve the total volume of available resources (staff and vehicles), efficient resource utilisation is critical to emergency service performance.

Using novel machine learning techniques, this project aims to develop methods that can more efficiently manage ambulance resources than human dispatchers. In collaboration with Wellington Free Ambulance, we learn from existing dispatcher expertise and years of historical dispatch data to improve response times to patients. In addition to response time, we learn dispatch policies which maximise paramedic break length, provide equitable service across the Wellington region, and are understandable to audit staff at Wellington Free Ambulance.

Primary contact: Associate Professor Yi Mei yi.mei@vuw.ac.nz

In 2019, the U.S. Federal Communications Commission estimated that improving the average ambulance response time by sixty seconds would save 10,120 lives per year, each valued at US$9.1 million. We infer that such an improvement projected to New Zealand could save up to 150 lives per year, each valued at $4.46 million by the Ministry of Transport.

The problem faced by ambulance services changes daily: they do not know how many emergencies of which urgency will occur when or where. New Zealand services act with a limited set of resources and struggle to meet the response time targets set by Government. While an increase in infrastructure funding would improve the total volume of available resources (staff and vehicles), efficient resource utilisation is critical to emergency service performance.

Using novel machine learning techniques, this project aims to develop methods that can more efficiently manage ambulance resources than human dispatchers. In collaboration with Wellington Free Ambulance, we learn from existing dispatcher expertise and years of historical dispatch data to improve response times to patients. In addition to response time, we learn dispatch policies which maximise paramedic break length, provide equitable service across the Wellington region, and are understandable to audit staff at Wellington Free Ambulance.

This programme aims to develop small-molecule drugs to prevent brain and motor damage in children with Krabbe disease (KD), a rare and fatal inherited disorder. Also known as globoid cell leukodystrophy, KD is caused by mutations in the GALC gene, which result in a deficiency of the lysosomal enzyme β-galactosylceramidase. This leads to the accumulation of toxic metabolite, psychosine, in the brain which damages nerve cells.

Children with KD appear healthy at birth but rapidly lose motor and cognitive function. Most affected children die before the age of two, with no treatment currently available.

Our team is using a powerful approach based on transition-state chemistry to design highly specific inhibitors of acid ceramidase (AC), a key enzyme involved in the production of psychosine. These inhibitors are designed to mimic the enzyme’s transition state, allowing them to bind with exceptional strength and selectivity.

This grant support the development of novel drug candidates, bringing us closer to a potential therapy for this devastating disease.

For more information, please contact:

Dr Farah Lamiable-Oulaidi

Team Lamiable-Oulaidi(external link) — Victoria University of New Zealand

This programme aims to treat Krabbe disease, an inherited disorder that results from the build-up of a toxic metabolite in the brain. This fatal neurodegenerative disorder, for which there is no treatment, is caused by a defective enzyme. We will use a unique technology known to deliver drugs that are highly efficacious and avoid side effects associated with off-target toxicity. As a result, we aim to deliver a life-saving drug candidate that will prevent irreversible brain and motor damage for children affected by Krabbe disease.

Honey bees contribute an estimated $5 billion to NZ’s primary industries. One of the greatest threats to the honey bee industry, here and internationally, is the parasitic varroa mite and the viral disease it spreads called Deformed wing virus (DWV). Together, varroa and DWV are the leading cause of death to honey bees worldwide. The current approach to controlling varroa and this virus is a chemical pesticide that is becoming ineffective.

Our research will develop a safe, effective and commercially viable method to control DWV and mitigate the effects of the varroa. Our method uses immunotherapy for bees. Immunoglobulin (IgY) antibodies have previously been developed to treat infections including influenza. The antibodies are cheap to produce and can be stored for long periods. This method of pest control also leaves no synthetic chemical residues in honey, so it has no ill effects for humans or bees. Our preliminary research indicates these antibodies have considerable promise.

We will develop a IgY antibody treatment and test its effects on the varroa parasite’s reproduction and fitness. Our treatment will be in a form that can be easily fed to bees by beekeepers. During field trials we will confirm that the antibody treatment is safe for bees and enhances their productivity.

Once our research is complete, we will work with the Ministry for Primary Industries and Environment Protection Agency to authorise the legal use of the treatment by beekeepers. We will then develop a pathway to commercially produce the new treatment.

Our goal is to develop an environmentally safe method to control this pest and disease in honey bees. However, this approach could become a model way to control many other pest species.

The recent finding that cells package genetic material into membrane-bound microvesicles for secretion has initiated a new era of biomarker discovery. This discovery has led to novel methods of disease detection that are either minimally- or non-invasive.

The embryo is no exception and microvesicles packed with genetic material and secreted into their surrounding environment provide a snap-shot of their genetic make- up. Our research will use this phenomenon to address a significant problem for the human fertility industry.

Half of human embryos generated by in vitro fertilisation (IVF) in fertility clinics have an incorrect number of chromosomes (aneuploid). The transfer of aneuploid embryos into the uterus results in embryo loss, which is emotionally and financially devastating to the recipients. Whilst an aneuploidy test is available which extracts cells from the embryo, it is invasive, risky to low quality embryos, expensive and has a long result turn-around time. These limitations mean very few people choose to get their embryos tested.

We will assess the microvesicle-encapsulated genetic material secreted from IVF-embryos to determine if they accurately indicate their chromosomal numbers. During this work, we will identify secreted biomarkers of specific aneuploidies for the first time and develop a simple, rapid and cheap test for their detection. The revolutionary advantage of this test is that it only tests the medium in which the embryos are cultured in, leaving the embryo undisturbed.

Such a test is highly desirable to the international fertility industry and we will work with industry partners and commercial genetic testing companies to develop a commercially-available test. The down-stream benefits of this non-invasive test is that more people will choose to get their embryos tested leading to an improved IVF success rate.

The Terahertz (THz) frequency range of the electromagnetic spectrum, sitting between infrared light and microwaves, has vast untapped potential for scientific, industrial and environmental uses - from detecting the evolution of galaxies to high bandwidth telecommunications and monitoring concentrations of atmospheric gases affected by climate change. The ability of THz waves to penetrate biomolecules and probe them without causing damage also makes them ideal for many areas critical to New Zealand such as agriculture, food production and biomedical imaging. However existing technologies for generating THz waves are severely limited in the range of frequencies they can produce and the instruments for doing so are bulky, expensive and little used outside of research labs. For that reason, this part of the spectrum has long been known as the 'THz gap', waiting for the technological advances that will finally open this underutilised region to its myriad beneficial uses.

We will develop a source of THz waves that covers the full range of frequencies in this spectrum, using a novel technique of generating THz from magnetic materials. Our new technology will overcome the limitations of existing sources and will lead to THz technologies that are affordable and suitable for use in industrial settings. We aim to stimulate the growth of an entirely new high-value and high-productivity industry in New Zealand based on the manufacture and use of THz technologies. Our goal is for New Zealand to become a global hub for THz technology R&D, manufacturing and services for current and future industries.

Phytophthora is a genus of microorganisms that cause devastating dieback and root-rot diseases in thousands of plants worldwide. The economic impact of Phytophthora diseases on crops and native ecosystems is billions of dollars per annum, and these impacts are predicted to worsen with climate change. Here in Aotearoa New Zealand, a recently identified species Phytophthora agathidicida is threatening kauri (Agathis australis), which are treasured, long-lived native conifers. Whereas another Phytophthora species (P. cinnamomi) causes root rot in key NZ crops such as avocados.

These pathogens are extremely difficult to control using existing agrichemicals, and the effectiveness of the few available treatments is jeopardized by increasing rates of resistance.

Using a bi-cultural approach, our team has identified naturally occurring compounds from native New Zealand plants that inhibit the growth and survival of Phytophthora pathogens (in the laboratory, at least!). Here, we will build upon this work – and explore how to take these results from the laboratory to the field. Our ultimate goal is to have formulated plant extracts that are safe, effective, and can be used to control Phytophthora diseases in our fields and forests.