Knowledge-based and Personalised Privacy Preservation over 5G and IoT Smart Devices. This project aims to investigate privacy preservation protocols in a 5G integrated IoT environment through an analysis of the depth of smart-device use in common smart domains. 5G’s addition to IoT-based smart devices will be effectively deployed and utilised by a large majority of individual and organisation-based users. The knowledge-based ontology and tools developed in the project will help form the new privacy preservation mechanisms that are required for the 5G enabled environment. The construction of new AI-based tools and testing facilities as well as the generation of new knowledge in the field of privacy preservation are expected outcomes of the study. This Project is supported by Australian Research Council’s Discovery Project grant.
Privacy-Aware Resource Optimization for Distributed AI in SMEs. This project aims to develop a secure and privacy-preserving AI system that enables many users to train large models without sharing their data. It expects to generate new knowledge in managing delays and interruptions in distributed training and in designing intelligent scheduling tools for efficient use of shared computing resources. Expected outcomes include scalable training algorithms, smart resource schedulers, and deployable prototypes supporting privacy-preserving AI services. This should provide significant benefits by making advanced AI more accessible and trustworthy for small and medium businesses, while reinforcing Australia’s leadership in digital capability, cybersecurity, and data privacy. This Project is supported by Australian Research Council’s Linkage Project grant.
Revolutionising Healthcare with Machine Unlearning. With the rise of artificial intelligence (AI) and machine learning (ML), healthcare is experiencing significant transformation, especially in patient monitoring systems that offer real-time health care. This theme represents a cutting-edge study on AI's role in these systems, highlighting advancements in federated learning, reinforcement learning, and the novel area of machine unlearning across settings like remote patient monitoring and mental health facilities. Findings underscore the power of AI in early anomaly detection and the innovative FedStack architecture's ability to provide individual insights. As the research tackles machine unlearning for data privacy, it acknowledges challenges in data scale and explainability, suggesting areas for future exploration. Overall, this theme will showcases AI's potential to reshape healthcare, emphasising its future prominence in patient care enhancement. Recent outcomes from this research theme includes the articles on top-tier journals like TKDE, TNNLS, and INFFUS.
Artificial Intelligence and Brain Informatics for Precision Psychiatry. The project aims to develop trustworthy AI methods that integrate multimodal data—including clinical records, neuroimaging, EEG, wearable sensors, and digital health data—to improve the diagnosis, prediction, treatment, and monitoring of mental health disorders. A particular focus is on developing predictive models to personalise repetitive transcranial magnetic stimulation (rTMS) treatment by identifying patients most likely to benefit, optimising treatment protocols, and monitoring therapeutic outcomes. The project combines machine learning, deep learning, multimodal foundation models, explainable AI, and privacy-preserving techniques to identify clinically meaningful biomarkers and support personalised psychiatric care. Through close collaboration with clinicians and healthcare providers, the research aims to deliver intelligent clinical decision support systems that enable earlier intervention, more effective treatment selection, and improved patient outcomes while ensuring the responsible, ethical, and secure use of AI in mental healthcare. This project is partnered with the Cannan Institute and Belmont Private Hospital, Brisbane.
Artificial Intelligence for P4 (Predictive, Preventative, Personalised and Participatory) Medicine. Recent successes in Biotechnology and Artificial Intelligence have been driving the transformation of medical practice from traditional untargeted, reactive and experience-based to targeted, proactive and evidence-based. P4 (Predictive, Preventative, Personalised and Participatory) medicine will provide cost-effective disease care, reduce the incidence of diseases and replicate the innovation cycle of systems medicine on a large scale, and is believed "a revolution of medicine / healthcare practice". This research is focused on predictive and personalised medicine by predicting potential diseases based on patient's personal health status using Machine Learning techniques. It will further support physicians' clinical decisions by providing prescription re-check and suggesting treatment plans using knowledge bases and information retrieval techniques. To achieve these goals, study of massive data in heterogeneous types is essential. The research will help develop our capability of proactive and evidence-based medicine and help design clinical decision support systems.
Computational Social Science for Online Mental Health using Artificial Intelligence. Many people are suffering from mental issues without knowledge of it. As a result, they are unable to access to appropriate helps. Finding and helping these people have motivated us in the research proposed in this project. It will model the behaviour of online social network users by analysing their expressions using natural language processing and machine learning techniques, and alert potential mental issues adopting data mining techniques like outlier detection. A knowledge base conceptualising mental health domain knowledge will provide foundation to these tasks. With the outcome of the work, clinical decision support systems can be designed to assist psychologists and social workers in diagnose and help people with mental issues at early stage; tools like mobile apps can be developed to help guardians like parents to keep an eye on their children's mental health proactively without breaching their privacy. People suffering from mental issues can also benefit from the tools by monitoring their own mental health easily, so that they could pull back at early stage and avoid falling into more severe circumstances if anything wrong is happening. The proposed research will make potential theoretical contributions to deepening our understandings of mental health, as well methodological contributions to knowledge engineering, natural language processing and data analytics.
