Health analytics
Data integration platform for patient-centred e-healthcare
The lack of medical data integration prohibits clinicians from building a comprehensive understanding of the patient condition and further deprives patients from developing a holistic awareness of their medical conditions.
In this project, we have designed and developed an open data integration platform for data concerning:
- patient data,
- clinical data,
- medical data and
- historical data
All data is siloed across multiple health information systems.
Our open platformcan accommodate and integrate further heterogeneous data sources such as data streams generated by wearable IoT devices. We have adapted our platform over time to address patient-centred healthcare and clinical decision support requirements at OrthoSport Victoria.
Collaborators:
- Dr. Kate E. Webster, Associate Professor, School of Allied Health, La Trobe University
- Dr. Brian Devitt, Orthopaedic Surgeon, OrthoSport Victoria
Online patient forum analysis
Online patient forums present a platform to seek, provide and exchange information in an inclusive environment. As the volume of text on patient forums continues to grow exponentially, it is important to improve the quality of retrieved information in terms of relevance, reliability and usefulness.
In this project, we developed a text mining approach that generates a knowledge extraction layer. This layer addresses the void in personalized information retrieval from online patient forums.
We have successfully applied our platform to a number of medical forums, providing insights into:
- undocumented symptoms,
- side effects and importantly,
- the emotional impact of varied medical conditions.
We are currently experimenting this approach with a team of clinicians from Eastern Health.
Collaborators:
- Dr. Weranja Ranasinghe, Urology Registrar at Austin Health
Using machine learning to identify individual differences in brain networks and recovering trajectories following stroke
Recovery rates for people who have suffered a stroke varies regardless of previous brain lesions in similar regions or capacity for rehabilitation.We will be constructing a learning model that can predict the best rehabilitation routine for stroke sufferers, based on:
- the patients’ MRI scan,
- their demographic background and
- other open-sourced information
Our goal is to aid those planning rehabilitation services with our learning model. For example, it may uncover new variables that a therapist has not considered previously when developing rehabilitation plans. It could also be used by junior therapists as a decision support system when delivering rehabilitation to patients.
Collaborators:
- Professor Leanne Carey, La Trobe University and Florey Institute
- Dr Peter Goddin, Florey Institute,
- Mr Alistair Walsh, La Trobe University
Functional MRI (fMRI) analysis for developmental disorders
We are exploring the application of novel machine learning and predictive analytics algorithms for fMRI analysis. This will allow us to distinguish between imaging from a control group and a test group.