Project Summary
This project will develop the first precision-based risk assessment tool to accurately identify individuals who are at risk of an adverse cardiovascular event. Using latest machine learning techniques and computation methods, the project will use a large retrospective data set with linked patient outcome data to develop and validate a unique tool which can be integrated into standard clinical practice via cloud-based computing software. Using the patient’s medical images, and other available health data, it will outperform current risk assessment methods. These current methods are based only on a few generic indicators and ignore potential patient-specific markers embedded within the available information. Specifically, by using latest computation methods, for the first time, we will be able to compare individual variation within the population. This will likely reveal the currently remaining 30% of unknown factors for cardiovascular health risk.
What is the issue for NSW?
In NSW, one out of three heart attack patients have been incorrectly identified to be not at risk and did therefore not receive standard primary care. This is because some risk factors for adverse cardiovascular events are still unknown today. Standard risk assessment is based on outdated risk scoring. This assessment is only calculated based on age, blood pressure, cholesterol levels and history of smoking. This shortfall in clinical care is a significant health care burden in NSW and globally. Being able to correctly identify patients at risk and therefore optimise treatment strategies would save many lives, improve well-being and save significant costs.
What does the research aim to do and how?
This research is proposing to use a computer and in-depth data driven risk assessment which will reveal previously unknown markers of risk and allow a rapid computed assessment during standard clinical assessment with medical imaging. The tool will be developed based on computation and linkage of health data outcomes in retrospective cases. Using this in-depth information, Dr Beier and her team will train an automated, rapid assessment tool using latest machine learning techniques. After development and training, this tool will be able to calculate risks in individuals within seconds using only the standard clinical information.