Student Projects
Developing a Decision Support System at ED triage for predicting health outcomes
Our previous two-year research project on Surgical Outcome Calculators examined surgical outcomes in NZ patients to help produce a surgical risk prediction tool: nzRISK, which produces a risk score for that patient at one month, one year and two years after surgery. NZ clinicians are now using the tool, which has produced tremendous interest and value.
Smart Patient Cohort Builder
The primary goal of this project is to provide a decision support system for triaging referrals to make this task more efficient.
Patients Like This
It is important to provide patients with reliable, accurate and personalised information about their health conditions, to engage them in their health. Often, this is done by doctors, based on their textbook knowledge and prior clinical experience in communicating with and managing similar patients under their direct care.
Triaging cardiologist referrals from GPs
Triaging GP referrals is a time-consuming task for doctors in New Zealand. It is common for doctors to spend more than ten hours per week on triaging electronic referrals to various risk categories or direct to investigation tests.
A Deep Learning Platform for GP Referral Triage
Heart disease is the leading cause of preventable mortality in Aotearoa, New Zealand. It is a disease that disproportionately affects certain groups such as Māori – who have higher rate of ischaemic heart disease (or coronary heart disease) and stroke but have lower access to health care.
Greater Engagement of Mental Wellbeing Solutions Through Personalisation
Mental health problems are a worldwide pandemic. Over 50-80% of New Zealanders will experience mental distress in their lifetime, costing an estimated $12B or 5% of GDP annually.
Telehealth as a Model for Health Care Delivery for Underserved Populations
In New Zealand, Telehealth provided health care during the COVID-19 lockdown, but its use has declined back to pre-COVID levels.
De-identification of GP e-referrals for a deep-learning based triage decision support tool
Medical student Nick James has developed a neural network model that can find identifying details in patient records, so the details can be removed making the records available for research purposes.
Deep learning-based Melanoma prediction from skin images
One method of detecting Melanoma is through the usage of machine learning on medical images.
Machine readable clinical guidelines and pathways
Machine readable clinical guidelines and pathways There has recently been rapid growth in the use of clinical guidelines and pathways for the assessment, diagnosis and management of medical conditions. Summer of Research Project by David Bassett, supervised by Dr....
Improved planning for the scheduling of surgeries
To better allocate resources and plan for surgeries, Counties Manukau District Health Board (CMDHB) require data-driven analysis of their surgical scheduling. This will help them to understand the effect of re-allocating surgical sessions between surgical specialties.
Survey of machine learning-based approaches to de-identification of medical documents
Protecting patient confidentiality while using medical data from electronic health records (EHRs) for research is crucial. De-identification of EHR data provides the opportunity to use the data for research without the risk of breaching patient privacy, and avoids the need for individual patient consent.
Using Council Data to Investigate Health Outcomes
Auckland Council has a range of geospatial and infrastructure data available, which could potentially be linked with health data to provide insights. The purpose of this project was to investigate the data held by the council and describe how it could be used to inform healthcare.
A Prediction Algorithm for Familial Hypercholesterolaemia in New Zealand
Familial Hypercholesterolaemia (FH) is a genetic condition which causes an increased risk of cardiovascular disease and premature death (3-4 times the risk of early death). It is estimated to affect 1:300 people in the general population of New Zealand.
A data mining project using the National Health and Nutrition Examination Survey dataset
Atrial fibrillation (AF) is the most common sustained heart rhythm disturbance. At present, 25% of the New Zealand population who are 40 years old or more will experience AF in their lifetime. AF increases morbidity and mortality.
Effects of resourcing on timely and effective care of patients in a lymphoedema clinic
Computer models can help design specialist health clinics, giving patients the better care and using clinic resources efficiently. Ellen Gibbs created a simulation for a clinic dedicated to treating lymphoedema.
An online system for answering medical questions
As demand for doctors increases, patients may be able to decrease pressure on the health system and save themselves money by getting a preliminary diagnosis from an automatic question-and-answer system.
Application of deep learning techniques in a de-identification system
Clinical data, collected by healthcare providers when treating patients, is incredibly valuable for medical research – but good data is hard to get.
Using modelling to improve efficiency at Dargaville Medical Centre
Creating a model of how patients and staff use and provide services at a health clinic can help that clinic operate more efficiently.
Financial Evaluation of EDDI (Early Detection Decision Information)
New software could flag issues early during operations, helping surgeons and anaesthetists intervene before serious complications arise. We need to find out how much of a difference this software makes, and whether it will prevent complications and save money.
Interpretable image-based machine learning models in healthcare
Neural networks can be great at solving problems, but they sometimes give wrong answers. Would you trust an algorithm that got life-saving information right 80% of the time?
Clinical abbreviations detection and normalisation
Efficient communication between healthcare professionals about a patient’s health is crucial to delivering the best possible care.
Evaluating Biomedical Word Embeddings
It takes time and money to train specialist medical algorithms, and medical-specific data is hard to obtain for research, but the payoff may well be worth it.
Feature importance for adverse drug event named entity recognition
Automatically identifying drug names, dosages and effects in health records could help researchers find relationships between certain medications and negative effects on patients.