PhD in Engineering

This PhD explores how Artificial Intelligence (AI) can make immersive dashboards more usable and effective for critical decision-making. Visual analytics dashboards are increasingly used to support decision-making in domains such as healthcare, finance, and education. However, they often suffer from information overload and poor usability, which raise cognitive load and increase the risk of errors. In safety-critical settings, even small mistakes can have serious consequences. The use of Extended Reality (XR) displays such as head-mounted virtual reality, mixed reality overlays, and room-scale CAVEs offers new ways to explore complex data. Immersive environments can make patterns and trends more visible, but if poorly designed they can worsen decision-making by overwhelming users with clutter, disrupting spatial awareness, and requiring awkward interactions when users' hands are occupied (e.g. handling equipment or speaking with patients). This PhD has two main goals: 1. Reduce information overload using Large Language Models (LLMs). You will investigate how models such as GPTs and LLaMA can translate natural language queries (e.g. 'Show vitals from the last 6 hours') into structured filters, and enhance their accuracy through ontology-aware prompting and domain-specific fine-tuning. 2. Enable natural voice-based interaction: You will develop a voice interface using an LLM-driven semantic parser, a validation layer to check parameters, and a dialogue system to resolve ambiguous queries (e.g. “Do you mean before or after ICU admission?”).