Artificial Intelligence is rapidly transforming healthcare by assisting clinicians in disease diagnosis, prognosis, and treatment planning. While recent advances in deep learning and large language models (LLMs) have significantly improved predictive performance, most existing AI systems remain passive prediction tools that lack transparency, reasoning capability, and reliability. These limitations hinder their adoption in real-world clinical practice, where explainability, trust, and accountability are essential.
Recent developments in Agentic AI offer a promising paradigm shift. Rather than relying on a single monolithic model, multiple specialised AI agents can collaborate to analyse patient data, retrieve supporting medical evidence, estimate uncertainty, verify each other's conclusions, and produce transparent clinical recommendations. Such systems more closely resemble the collaborative workflow of multidisciplinary clinical teams and have the potential to provide more reliable and trustworthy decision support.
This PhD project aims to develop next-generation Trustworthy Agentic AI frameworks that integrate multi-agent collaboration, large language models, retrieval-augmented generation (RAG), explainable AI, and uncertainty estimation to support evidence-based clinical decision making.
Required knowledge
- Python programming
- Machine learning and deep learning fundamentals
- Large Language Models (LLMs) and Generative AI
- Multi-agent AI or AI reasoning (desirable)
- Explainable AI (XAI) and trustworthy AI concepts
- Basic knowledge of biomedical AI or medical imaging (desirable)
