Research Themes
My research focuses on trustworthy multimodal AI for mental health, especially speech-based depression detection.
Multimodal Speech-Based Depression Detection
Audio and text-based modelling for detecting depression-related patterns from clinical interview speech.
Federated Multimodal Learning
Privacy-performance optimisation for healthcare AI where sensitive data cannot be centrally shared.
Privacy-Performance Trade-offs Analysis
Quantifying the trade-off between predictive performance and privacy preservation to support trustworthy federated AI for healthcare.
Representation-Centric Interpretability
Moving beyond prediction-only explanations to inspect model representations and clinical alignment.
Responsible Translation
Exploring safety, limitations, privacy, and clinical usefulness of AI-based mental health screening.
Beyond Depression
Extending the applicability of multimodal AI models to other mental health conditions and populations.