Healthy Society

Discover how our students are shaping a healthier society. 

Researchers creating prosthetic hands
  • Endocrine disruptor mechanisms

    Understanding the molecular mechanism of endocrine disruptors.

    Project by: Victoria Severin 

  • AI to characterise REM sleep EEG

    Accurate monitoring of REM sleep substages is critical for understanding disease progression in conditions such as Parkinson's disease, insomnia, and PTSD. However, current methods are often complex and impractical for routine use. There is a clear need for a simpler approach that can reliably identify REM sleep patterns using minimally invasive data. This project addresses that need by applying machine learning techniques to electroencephalography (EEG) recordings to explore natural groupings within REM epochs, corresponding to the two recognised substages. By uncovering patterns and features that differentiate these substages, we have laid the groundwork for developing an automated system for REM substage classification. While full classification has not yet been implemented, this exploratory analysis provides key insights into the EEG signatures of REM substages, paving the way for accessible, non-invasive monitoring that could inform treatment decisions and improve patient outcomes.

    Project by: Matthew Larsson

  • Malaria transfection plasmid checks

    Malaria is a mosquito-borne parasite that results in more than 600,000 deaths per year, largely in tropical/sub-tropical regions, and currently there is no vaccine. The function of 30% of malarial proteins are unknown, thus the best target for a malarial vaccine may not be known to us. One way to investigate protein function is to compare the localisation of the target protein to that of a known organelle marker in the cell, with two different colour fluorescence tags enabling localisation detection and discrimination between the two. Using Escherichia coli to propagate the plasmid, restriction enzyme digestion and sequencing of the plasmid DNA was used to verify the sequence of the DNA plasmid. Upon this confirmation of the plasmid sequence, the plasmid will be used in future studies to identify proteins to prioritise as vaccine candidates.

    Project by: James Maletto 

  • Genomic autopsy incidental findings

    In Australia, approximately 1% of pregnancies result in pregnancy loss, termination, stillbirth, or neonatal death, with 285 miscarriages and 8 perinatal deaths occurring every single day. Performing a genomic autopsy alongside standard-of-care can enable a diagnosis in 2 out of 5 cases, and provide clarity during times of bereavement. While analysing genetic data to investigate the underlying cause of pregnancy loss or perinatal death, there is the possibility of identifying susceptibility to unrelated genetic disorders in the parents. There is an ethical obligation to report incidental findings, which are usually medically actionable or require clinical management. There is a need to perform systematic analysis of secondary findings to examine the rate at which they are present in this cohort. These findings have the capacity to increase awareness and facilitate immediate treatment or early monitoring for late-onset disorders, ameliorating their impact.

    Project by: Amelia Vlahakis 

  • Genetic variation in hops flowering

    Hops are one of the key ingredients used in making beer, as a natural preservative and bitterness flavouring agent. Within Australia, growth primarily occurs in the low latitude regions of Tasmania and southern Victoria, due to the day length, or photoperiod, required to trigger flowering. If this dependence on photoperiod can be overcome, there will be opportunity for the industry to expand into other areas of Australia. This project is aiming to identify any differences in the genome of commercially grown hops, in the alleles suspected to be involved with photoperiod-regulated flowering.

    Project by: Ruby Schwarz 

  • Physical model of spinal fluid flow

    Have you ever wondered what happens when the vital fluid, cerebrospinal fluid (CSF), that flows around your spinal cord is blocked? We are studying how spinal stenosis – a narrowing of the fluid passage around the spinal cord – and the flexibility of spinal tissues, affect the flow of CSF. We have developed a flexible flow phantom – a physical model – of the pig spine to help us in our study. Pigs are commonly used for spinal cord injury (SCI) research due to their anatomical similarity to humans. The physical model will provide insight into the interactions between CSF flow, spinal stenosis, and tissue flexibility. The outcome of this project will support ongoing research into biomarkers related to CSF dynamics following SCI.

    Project by: 

    • Govind Kumar Seewsagur 
    • Aryan Katyal 
    • Jayden Lee Iuliano 
  • Algorithms to locate atheroma in 3D

    Currently, dual modality imaging using 2D near-infrared fluorescence (NIRF) and 3D optical coherence tomography (OCT) is used in atherosclerosis diagnosis. By developing a 3D reconstruction of the NIRF images, clinicians will be more informed when assessing patient conditions. The goal of this project has been to develop an algorithm that predicts the 3D fluorescence from an atheroma; this is possible through the generation of light transmission models of the artery using ray tracing from OCT data. The light transmission models generated determine the light absorbed and emitted from bodies. The algorithm is tested using simulated data in a lower dimensionality 2D sandbox to predict the fluorescence of bodies and comparing this to the ground truth. The current standing in lower dimensionalities show promising results with further testing to improve the algorithm, with further work is being completed to move the system into full dimensionality.

    Project by: 

    • Angus Heath 
    • Jack McAuliffe 
  • Treating oxidative stress disorders

    Oxidative stress is associated with a build-up of reactive oxygen species (ROS) inside cells and leads to the onset and progression of several diseases, including cardiovascular disease, neuropathy and neurological diseases such as multiple sclerosis. Monomethyl fumarate (MMF) is a drug designed to treat oxidative stress. However, its systemic action throughout the body results in numerous side effects, including immunosuppression, gastric ulceration and kidney damage. This limits its usefulness to treating debilitating neurodegenerative disorders. To overcome these limitations, a MMF prodrug has been developed at the University of Adelaide. The prodrug remains inactive until it encounters high levels of ROS, at which point it reacts and selectively releases MMF at the site of oxidative stress. The prodrug was found to have reduced side effects and reverse neuropathic pain in mice. Enhancing the prodrug's activity while reducing its side-effects further is key to finding an effective treatment for oxidative stress disorders.

    Project by: Cormac Baker

  • Selection model for proton therapy

    With proton beam therapy (PBT) set to come to Australia, there is a need for a broad selection process to filter through patients requiring cancer treatment. Studies have shown that in the case of left sided breast cancer, a small number of patients may benefit from PBT. This project aims to create an automated framework to refer patients with left sided breast cancer to PBT comparison planning centres. Radiation delivered to the heart during conventional photon therapy is used to estimate a relative increased risk of coronary events. This risk is compared to a threshold to determine if PBT could be beneficial. The model will identify patients to be referred for a PBT comparative plan. This model will serve as a proof-of-concept for a system to be implemented in Australia and will be the first implementation of automated model-based pre-selection worldwide.

    Project by: Mary Jones

  • Bile-taming bugs to prevent cancer

    Colorectal cancer is one of the commonest and deadliest cancers worldwide, and in Australia, cases in people under 50 have more than doubled in the past two decades. Recent research has emphasised the strong links between gut bacteria (the microbiome) and bowel diseases like cancer. In particular, patients with colorectal cancer often have gut bacteria which modify bile acids – substances that help digest fats – into more toxic forms. These toxic bile acids may upset the balance of the gut microbiome, increase inflammation, and contribute to cancer development. This project explored a new idea: can we reduce cancer risk by altering bile acids in the gut? As a first step, we looked at whether we could engineer the common probiotic bacterium, E. coli Nissle, to produce bile-acid modifying enzymes. In doing so, we assessed the initial feasibility of a microbiome-based colorectal cancer prevention strategy.

    Project by: Jasmine Esvelt

  • Creating anti-cancer nanoparticles

    The biggest challenges in medicine might one day be solved by some of the smallest solutions.

    This project explored how to create tiny particles, called nanoparticles, that can carry high amounts of cancer medicine directly to tumours. Traditional chemotherapy often spreads medicine throughout the whole body, which reduces its effectiveness and causes side effects. By contrast, nanoparticles act like miniature delivery vehicles that can hold more medicine and stay stable until they reach their target.

    The project focused on making these nanoparticles and testing their key properties. Special tools were used to measure their size, how stable they remained during storage, and how much drug they could hold.

    The results showed clear guidelines for producing nanoparticles that are both stable and capable of carrying large amounts of medicine. These findings form an important foundation for improving cancer treatments in the future.

    Project by: Yash Patil

  • MRI-derived 3D-printed prosthetics

    For amputees, a well-engineered prosthesis can bridge the gap between limb loss and the freedom of mobility. Over 25% of amputees choose to abandon their prostheses due to intolerable discomfort from incorrect load distribution and insufficient adaptability to daily limb volume fluctuations. The socket is the interface between the amputated limb and the prosthetic, serving as a fundamental component of the human device system that dictates comfort and usability. In this project, we used a systems engineering framework to develop and verify a transtibial prosthetic socket prototype from MRI data incorporating 3D printing and pneumatic volume control. These techniques enabled precise anatomical mapping of the residual limb, efficient fabrication of complex geometries, and dynamic adjustment for residual limb volume changes. Computational modelling and physical testing with our amputee stakeholder yielded a bespoke design that provided optimal load distribution, comfort, stability and scalability for broader clinical application.

    Project by:

    • Bailey Giles 
    • Taylor Sarich 
    • Taner Herkiloglu 
    • Jack Higgins 
    • Josiah Halkias 
  • Tiny tumours, tailored treatments

    What if tiny, lab-grown replicas of a patient’s own gastric cancer (GC) tumour could help create personalised and more effective treatment plans for the world’s fourth deadliest cancer? This is the potential of gastric organoids – miniature stomachs grown from a patient’s tumour – which were found to closely mimic a patient's response to cancer treatments.

    A challenge in using organoids to determine treatment plans lies in cancer’s aggressive progression, demanding rapid testing and commencement of treatment. Current research focuses on establishing and testing organoids to ensure they can be used within clinically relevant timeframes.

    In this experiment, tumour organoids were grown from a HER2 positive and negative GC patients’ tumours and treated with Lapatinib alone and in combination with FOLFOX. GC markers and genetic mutations allow analysis of organoids and corresponding tumour tissue’s similarity. The results will provide insight into how HER2 positive GC patients might respond to this targeted therapy.

    Project by: Stephanie Jukic 

  • Cheap wireless sensing technology

    What if the air around us could speak? From farms to factories, invisible VOCs (Volatile Organic Compounds) remain undetected until they become dangerous. Traditional VOC detectors are bulky, costly and require constant maintenance. These detectors became inaccessible to many small businesses and communities.

    Our project explored two different sensing approaches. The first one is to develop a low cost, wireless alternative by turning everyday RFID tags into tiny, battery-free “noses”. By combining chemical-sensitive materials with the tags, we enable them to sense hazardous VOC and send data through radio waves. The second one is a microcontroller-based system that directly measures VOC concentrations through sensor arrays.

    Both systems will be tested to determine which will work best for detecting different types of VOC. This technology could help small factories, farms, and other industries to afford VOC detection systems that can keep the safety of their workers without worrying about the budget. 

    Project by: 

    • William Thendean 
    • Cong Khoa Nguyen 
    • Fauzan Syahmi Bin Shapi Ei 
    • Siyu Zhou 
  • Noise reducing motorcycle helmets

    Travelling at high speeds leaves motorcyclists exposed to high levels of wind noise, which can cause hearing damage and reduce rider safety. Our project aims to reduce this noise by developing passive noise-reducing modifications for motorcycle helmets, without increasing drag or reducing safety. To achieve this, we combined computer simulations with physical testing, comparing both modified and unmodified helmets. Testing was performed using an acoustic head model fitted with microphones to replicate human hearing. The helmets were evaluated both on-road, where the head was mounted to a vehicle to replicate real riding conditions, and in a wind tunnel for more controlled aerodynamic testing measuring both wind noise and drag forces. By combining simulation and physical testing, the design and location of the helmet modifications have been optimised to have the greatest reduction in the effects of wind noise affecting motorcyclists, without compromising safety.

    Project by: 

    • Patrick McGavin 
    • James Wheeler 
    • Milo McNicholas 
    • Noah Osmond
  • Improving sleep with AI-made data

    Poor sleep quality can lead to serious health problems. To assess the sleep quality of a patient, doctors and experts record the brainwave of the patient and detect irregular patterns in the brainwave which relates to sleep instability. In this era of Artificial Intelligence (AI), there have been software systems capable of reading the brainwave and learning such irregular patterns to detect sleep instability. Nevertheless, these systems are still not perfect; the accuracy is not good enough for mass deployment. One of the approaches to improve these systems is to provide them with more data of brainwave. However, recording real brainwave data is a tough and time-consuming process. In this project, we utilise AI to generate synthetic data of brainwave that would be used in sleep-instability-detection systems. The synthetic data will help improve the accuracy of these existing systems while reduce the need of recording real sleep data.

    Project by: Andrew Luu 

  • Automated non-chemical fungicide

    Imagine if plants could get sick just like people do. In vineyards, grapes often catch a disease called powdery mildew, which looks like white dust on the leaves and ruins the fruit. Right now, farmers fight it with excessive chemical sprays, but those sprays aren't great for people, the planet, or the grapes themselves.

    Our project sets out to find a cleaner, safer way to treat this disease. Instead of chemicals, we used a special type of ultraviolet light called UVC. This light can kill the mildew without damaging the plant. To make it work, we designed an automated system that moves the lights up and down, in and out, aiming them exactly where the grapevine needs treatment, such as under the leaves where mildew tends to grow.

    The outcome was a design for an automated system that could provide grapes strong protection, reduce chemical use, and help vineyards grow healthier, tastier fruit.

    Project by: 

    • Andy Le 
    • Cian Dunlevy 
    • Jaedh Rameezdeen 
    • Vedant Sakpal 
    • Milan Tamang 
  • Heart disease can give you maths

    Atherosclerosis, a disease that involves plaque growth in the body's arteries, is the main cause of heart attacks in the world. Calcium's role in plaque development is still a mystery, even though research suggests that it is vital in this process. The coronary arterial calcification score is an accurate way to test how at risk someone is of plaque rupture.

    We want to mathematically model the role of initial calcification in atherosclerosis over time. Research suggests that calcification is sometimes helpful and sometimes not. The model aims to provide clarity on the effects that calcification has on the plaque.

    We use various mathematical techniques to create and explore our model, focusing on important variables in calcification. We investigate the stability of variables and any important points that the model reach, interpreting what they mean physically. By modifying the parameters in the model, we can investigate different biological scenarios.

    Project by: Faith Sawers 

  • Safer, steadier heart procedures

    Heart disease is one of the leading causes of death worldwide, and many patients require delicate procedures where doctors guide a thin tube, called a catheter, into the heart. These life-saving procedures can be challenging as the catheter may move, making it harder for doctors to work with precision. This project aims to design a support structure to stabilise the catheter, like giving the surgeon an extra hand during the operation. This project is currently in the design phase, with a focus on exploring different shapes and materials to find a solution that is safe, practical, and suitable for hospital use. The ultimate goal is to develop a stabiliser that can be sterilised and reused in clinical settings. While the final product is still in development, the project has the potential to make heart procedures safer, more efficient, and more reliable for both patients and doctors. 

    Project by: Olivia Folino Gallo 

  • Helping treat cancer with plants

    Parthenolide is a compound derived from Tanacetum parthenium (feverfew) with pharmaceutical applications such as cancer treatment. After extraction from the feverfew plant, using ethanol or solvents, the crude extracts contain many undesirable impurities in addition to the target parthenolide. These impurities need to be removed to achieve the required degree of product purity. This can be accomplished through serial tangential flow filtration (TFF), which is implemented as an additional downstream processing stage to remove these impurities, such as undesirable proteins and peptides, whilst allowing the smaller sized parthenolide to pass through the filters. This project investigated different operating conditions for the extraction of parthenolide from plant biomass, such as wet versus dry feed, and the optimum filter sizes from 0.1 µm down to 1kDA. Over the course of four months, an optimised procedure was developed for extraction of parthenolide which resulted in a substantial reduction of impurities.

    Project by: 

    • Kalarni Leonard-Down 
    • Ethan Wilkinson 
  • Non-invasive ID of SA microbats

    Microbats in Southern Australia provide ecosystem services, including pest control for South Australia's $1.81 billion wine region. However, many Australian bat species, such as the Southern bent-wing bat, are threatened due to anthropogenic impacts such as climate change and habitat loss. This project involves extracting DNA from unknown microbat scat samples from vineyards and Southern bent-wing bat scat samples from roosting caves. To determine 1) the association between South Australian vineyard locations and bat species visiting, 2)the sex bias in bats visiting South Australian vineyards, and 3) the sex bias in Southern bent-wing bat populations. Understanding these biases and associations will aid in encouraging microbats to places where they can act as biological controls and help guide conservation management of threatened microbat species.

    Project by: Emily Davies 

  • A clearer view inside the heart

    Cardiovascular disease is the leading cause of death in Australia, caused by fatty build-up inside arteries that restricts blood flow. To guide treatment, clinicians use an imaging tool called optical coherence tomography (OCT), that uses light to image the layers within an arterial wall. However, OCT probe spin at high speeds, and any irregular movement can blur the images. This blurring makes it harder for clinicians to accurately assess dangerous artery blockages.

    This project develops a standardised testing framework to overcome this challenge. We first use computer simulations to design probes that minimise image distortion. A phantom linear stage is then used to verify the resolution performance of these probe lens at high imaging speeds. Finally, the probes are tested in artificial artery models that mimic real human vessels, with fluid flow and heart beat phases.

    This framework will advance OCT technology, supporting earlier diagnosis, preventative treatment, and stronger clinical decision-making for cardiovascular disease.

    Project by: 

    • James Choimes 
    • Kiran Manchery 
    • Binil Scaria 
    • Benjamin Cullum 
    • Elijah Barrott-Walsh 
    • Hamad Alhammadi 
  • Bad sleep - is it in your genes?

    Obstructive sleep apnoea (OSA) is a common sleep disorder affecting about 8% of Australian adults and is highly heritable, with genetics explaining around 40% of risk. OSA causes repeated pauses in breathing during sleep, leading to disrupted rest, excessive daytime sleepiness, and increased risk of cardiovascular and metabolic complications. This project explores whether genes known to influence traits like bmi, blood pressure, glucose, and cholesterol also contribute to OSA. Using data from the Western Australian Sleep Health Study (WASHS), a cohort of over 4,000 participants with detailed clinical and genetic information, a Mendelian randomisation approach will test whether genetic risk for these traits/phenotypes directly increases OSA risk or acts indirectly as a result of the phenotypes themselves. Uncovering genetic pathways influencing OSA can inform tailored prevention strategies and interventions, helping clinicians predict risk and design personalised approaches to improve sleep health and overall wellbeing.

    Project by: Alexander Dawson

  • Next-gen dental surgery models

    The future of dental surgery training is shifting away from outdated cadaver-based methods, which are expensive, tightly regulated, inconsistent, and can even pose health risks. Our project focuses on developing an artificial training model for the removal of wisdom teeth, in collaboration with biomedical company Fusetec, using advanced 3D-printing technology.

    We begin by creating a detailed 3D model from medical CT scans and oral anatomy data. Careful material selection ensures the printed structures mimic the strength and feel of real teeth, gums, and bone, while also working within the capabilities of Fusetec’s printers. The model is produced using MultiJet Printing, which allows multiple materials to be combined in one print.

    Each prototype undergoes physical testing and evaluation by experienced dentists to assess accuracy and realism. Feedback from these tests drives improvements in design. So far, we have completed CT scan analysis and early 3D modelling, with material testing and full model validation to follow.

    Project by: 

    • Lachlan Grigg 
    • Cooper Pyman 
    • Ethan Schwerdt 
  • Soil-friendly mineral magnets

    Crops need small amounts of minerals like iron, manganese, and zinc for healthy growth, but in many soils these nutrients become trapped and unusable. Farmers often add “chelating agents”: compounds that hold onto minerals and keep them plant-available. A common one, EDTA, works well but remains in the environment for decades and can release harmful metals. LSA is a more eco-friendly option but is less effective at keeping minerals plant-available.

    This project seeks a better balance. We use pyroligneous acid (PA), a liquid made from waste wood, and treat it using ozone to remove undesired compounds. By adjusting ozone-treatment conditions, we aim to produce an effective, biodegradable compound that ensures healthy plant growth. The effect of adjusting these conditions on iron, manganese, and zinc availability and stability will be tested in soil. The outcome will be a green option that effectively supports crop growth while reducing environmental harm.

    Project by: 

    • Angelina Vu 
    • Daniel Perkins 
    • Quang Thai Tran 
  • AI for early stroke detection

    Every year, millions of people around the world suffer from strokes, a serious medical condition that can lead to long-term disability or even death. The key to saving lives is early detection and prevention. Our project explores how computer intelligence can help doctors predict who might be at risk of having a stroke before it happens. By studying health information like age, lifestyle, and medical history, we teach the computer to find hidden patterns that humans might miss. The outcome is a smarter tool that can warn people and doctors earlier, helping them take action to stay healthy and safe. In the future, this technology could support hospitals and clinics in preventing strokes, improving patient care, and reducing the burden on families and healthcare systems.

    Project by: Jialong Zhu

  • Modelling dry hopping in beer

    Dry hopping is an additional step to the brewing process to enhance the flavour and aroma profile of beer. It contributes intense aroma profiles like citrus, floral and tropical, without increasing the bitterness of the beer. Scaling up to industrial size batches from smaller sizes is difficult as the mass transfer of hop components are different in batch sizes. Our aim is to simplify the scale-up process by establishing a digital twin of the entire brewing process, which can predict the quality and production rate at different points during production. To develop this twin model, three 20 litre batches of pale ale were brewed with different amounts of dry hops. The digital twin was then validated with a larger scale at 200 litres. This model enables more consistent, efficient and flavour-optimised brewing.

    Project by: 

    • Andrea Dancel 
    • Olivia Hobby 
  • AI to catch stroke risk early

    Every year, millions of people suffer from strokes, which can result in death or permanent disability. The biggest challenge is that strokes often come without warning. My project asks a simple but vital question: can we identify the risk earlier?

    To accomplish this, I collected regular health information such as age, blood pressure, and lifestyle. I then used computer "learning" software, which hunts for patterns in the data that are not obvious—similar to how your brain learns from experience. This allowed me to see which factors are most strongly linked to stroke and to test whether a computer could give an early warning sign.

    The outcome was heartening: when the data was cleaned and laboriously balanced, the computer was better at identifying people who were at higher risk. One day, the technique could give physicians a straightforward "early warning" system—preventing strokes before they happen.

    Project by: Prasad Torane 

  • Cutting-edge dental ceramics

    Dental ceramics are essential for crowns because they closely resemble natural teeth and are highly durable. Zirconia, in particular, has become a leading material of interest due to its strength and aesthetic qualities. However, current manufacturing processes can introduce cracks, surface roughness, and hidden defects that reduce the long-term reliability of restorations.

    To address this challenge, this project will explore new digital manufacturing techniques for shaping zirconia. The project will investigate how different machining parameters affect surface and subsurface quality using advanced imaging tools, and test durability through fatigue simulations.

    By identifying manufacturing methods that minimise damage and improve surface finish, this research aims to provide clearer guidelines for dental laboratories. The long-term goal is to produce zirconia restorations that last longer, perform more reliably, and ultimately improve patient comfort and confidence.

    Project by: 

    • Daniel Chebotenko 
    • Talitha Owies 
  • The artificial spinal cord

    More than 20 million people currently suffer from a spinal cord injury, severely reducing quality of life, yet no effective treatments exist. Artificial spinal cords are a promising medical advancement for restoring spinal functionality through regeneration, replacement, and improved surgical training, with non-swelling hydrogels emerging as a particularly suitable material. This project aims to investigate and optimise hydrogel composition to closely replicate the strength and flexibility of the human spinal cord. Various hydrogel formulations were mechanically tested under different loading conditions, and swelling capacity was assessed. Based on these results, a statistical model was developed to predict hydrogel properties according to composition. This work demonstrates how the mechanical properties of hydrogels can be modified, providing valuable insight into the tunability of hydrogels for spinal cord applications. Overall, this research contributes to the development of surrogate spinal cords, supporting improved surgical training methods and laying groundwork for future implantation.

    Project by: Caitlin Edwards

  • Preventing West Nile virus

    West Nile virus infects thousands of people worldwide each year. It is transmitted by mosquitos and is expected to increase as climate change expands mosquito habitat. Luckily, most infections are asymptomatic or result in mild-flu like symptoms - however in some people infection can lead to encephalitis (swelling of the brain) and possibly death. This is more likely in those that have a weakened immune system. In order to develop antivirals and protect people from West Nile virus, we must first better understand the virus and how it causes disease. This project focusses on determining whether specific genes in the immune system assist or hinder during infection with West Nile virus. These findings contribute to the total understanding of the virus, and can be used in the future to identify candidates for drug development, helping to prevent serious illness in a changing climate.

    Project by: Albert Thompson 

  • Breaking bio: nano edition

    What if cancer drugs could wear disguises?

    Chemotherapy is powerful, but it often harms healthy cells, triggers unwanted immune responses, and doesn’t last long in the body. This project aimed to create a smarter, more targeted way to treat colorectal cancer using a hybrid system. This system has two parts: a lipid-nanoparticle core that carry the drug, and a coating made from cancer cell membranes. The coating helps the particles hide from the immune system and find the right cancer cells to attack. We made these “undercover” particles by combining lipid nanoparticles with membranes from HT29 cancer cells. Then we tested their size, charge, protein content, and how stable they were. The results showed that our disguised particles stayed stable in blood and during storage whilst retaining the proteins needed to attack cancer cells. This suggests that the system could be a more precise and effective way to treat cancer.

    Project by: Preshna Nakarmi 

  • Evolving smarter AI models

    Whether it's voice assistants or medical imaging, many of these tools are powered by Artificial Intelligence (AI). However, creating the "brains" behind AI, known as neural networks, is time consuming and often results in systems that are accurate, yet very large, slow and energy intensive. This means they are not allowed in a real world application where resources are limited.

    Our project investigates how smarter and more efficient artificial intelligent systems can be automatically designed by combining two biologically inspired strategies: evolution and self-adaptation. Using an evolutionary algorithm, we get the AI models to "compete" with each other, retaining only the best designs but improving them over generations. At the same time, each model can tune its own learning behavior, so the system can get around wasted work and find better solutions more quickly.

    This research helps to build AI that is practical, sustainable and more suitable to real-life applications.

    Project by: Vedansh Kumar 

  • Mapping the human journey

    Wallacea is an archipelago connecting Asia and the Sahul region that has undergone thousands of years of sea and glacial fluctuation. This history provided the perfect conditions for migration across briefly accessible corridors of land, leading to waves of ancient human movement beginning around 50,000 years ago followed by subsequent waves in the last 10,000-4,000 years. The dispersal of early humans is an important area of research as it allows for a deeper understanding of early population dynamics and history and the possible connections to disease and localised evolution.

    A method commonly used to investigate migratory patterns is the sequencing of mitochondrial DNA, which is a largely unchanged source of DNA in our cells that provides insight into maternal lineage. Within my project, I used pre-sequenced data from the mitochondrial haplogroup 'M20' to investigate lineage and migration history in the Wallacean region and later Austronesian expansion.

    Project by: Arabella Hawking 

  • Unravelling DNA complexes

    Many diseases are caused by errors in our DNA. Some diseases can be treated by stopping a corrupted section of DNA from functioning. This can be done by inserting another chemical strand into the DNA structure to form what is known as a DNA triplex. As part of this, scientists must know whether the extra strand has been correctly inserted. There are several ways to do this but there is concern within the scientific community that different methods of creating DNA triplexes and different methods of analysing these structures produce inconsistent results. This project compares the analytical utility of mass spectrometry, UV-Vis spectroscopy and gel electrophoresis for modified DNA structures. While all techniques can analyse DNA triplexes, the results can differ depending on the formation conditions and the analytical technique used. This work is expected to have ramifications for how DNA modification is evaluated and reported in the literature.

    Project by: Luke MacKay 

  • Drugging the undruggable

    Proteins are essential for all life, they are the workers inside our cells, each with a specific job which helps keep our body running. Despite their importance, there is still much we don’t know about their behavior. One way to study this is through loss-of-function, where we “switch off” certain proteins in cells to study them. Drugs that directly target proteins are incredible useful in this process, but many proteins are considered “undruggable,” due to their unique molecular properties. To overcome this, we have applied a cutting-edge degron technology, which allows us to tag proteins with a druggable interfaces for loss-of-function studies. Here, we apply this tool to one of the most difficult to target protein groups: transcription factors, which control our genes. By investigating p53, a transcription factor that is the most mutated gene in cancer, we are sharpening our understanding and demonstrating the value of these novel tools.

    Project by: Kasia Coultas

  • Teaching cells to smell

    Cancer is one of the leading causes of death worldwide, and while a treatment called CAR T cell therapy has shown remarkable success in blood cancers, even in patients deemed to be untreatable, it has not been successfully translated to solid tumours. CAR T cells are immune cells, harvested from the patient and genetically engineered to have a special "antenna" that helps them recognise and kill cancer cells. However, one major problem is that CAR T cells often cannot find or enter the tumour environment. This project aims to change this by giving them better navigation tools to follow the chemical "scents" released by tumours, called chemokines. Using a CRISPR activation system, we are testing whether we can switch on these chemokine "detectors" in model T cells. If successful, this could be an important step towards the next generation of cancer therapies that can infiltrate and destroy solid tumours.

    Project by: Rory Bright

  • Deep learning for better sleep

    Sleep is vital for our health, but many people suffer from hidden sleep problems that are hard to detect. One important signal doctors look for is called the Cyclic Alternating Pattern (CAP). CAP shows up in brain activity during sleep and can reveal issues such as insomnia, epilepsy, and sleep-disordered breathing. Right now, experts must watch hours of recordings to mark these patterns by hand, which takes a lot of time and can vary between people.

    This project uses artificial intelligence methods of deep learning to automatically find these sleep patterns. By teaching a computer to “learn” from real brainwave recordings, we aim to detect CAP events more quickly and accurately than current methods. This approach removes the need for manual analysis and could help doctors better diagnose sleep disorders. In the future, our system could be integrated into clinical tools to improve patient care and advance sleep research.

    Project by: Krunal Tushar Patel

  • Decoding sleep with deep learning

    Understanding sleep is essential for keeping our brains and bodies healthy. Conditions like teeth grinding, insomnia and breathing problems during sleep can affect how well we sleep. One way to study sleep is by looking at patterns in brain activity called Cyclic Alternating Patterns. These patterns reveal important information about sleep quality. However, traditionally detecting them from brainwave signals has been complex, time-consuming and requires expert knowledge.

    This project focusses on making CAP detection faster, easier and fully automatic using artificial intelligence by combining Convolutional Neural Networks(CNN) and Recurrent Neural Networks (RNN) including Long Short-Term Memory (LSTM) networks. By improving the way these models work the aim is to make a more accurate and reliable method to study sleep. Ultimately, this can help researchers and healthcare professionals better understand sleep patterns and their impact on brain and overall health.

    Project by: Karanbir Singh 

  • Optimising genome assembly

    Current genome assemblies – the foundational datasets enabling all genomics research – contain critical errors that undermine biological discoveries. Sea snakes represent one of evolution's most dramatic adaptive radiations, quickly transitioning from land to marine environments, yet flawed genome assemblies have obscured the genetic mechanisms driving this remarkable adaptation.

    Through systematic computational analysis, we identified systematic misassemblies between chromosomes 2 and Z across multiple sea snake species in publicly available genomes. These errors artificially separate chemosensory receptor genes from their regulatory elements, creating false evolutionary narratives. Using advanced bioinformatics pipelines, we reconstructed high-quality chromosome-scale assemblies that revealed the true genomic architecture.

    Our corrected assemblies uncovered structural variations in vomeronasal gene clusters – sensory genes crucial for detecting chemical cues underwater. Enhanced gene counts and accurate regulatory mapping demonstrate how chromosomal rearrangements enabled sea snakes to adapt their terrestrial chemosensory systems for marine life. This computational framework establishes new standards for genome quality assessment, ensuring reliable foundations for evolutionary research and providing essential tools for understanding adaptive radiation across species.

    Project by: Billy Trim