Research

My research examines how artificial-intelligence systems can support complex decision-making while remaining transparent, robust, fair and operationally useful. The work sits at the intersection of AI, Information Systems, data analytics and human-centred computing.

Explainable, trustworthy and fair AI

I work on methods for evaluating and improving explanations produced by machine-learning systems. This includes neighbourhood-based XAI, sufficiency and necessity, explanation utility, fairness evaluation, bias detection and the design of interpretable decision-support workflows. A central question is not simply whether an explanation can be generated, but whether it is useful for the specific decision or explanandum being addressed.

Generative AI and model evaluation

A second strand examines the quality, diversity and reliability of generative models. My work has considered mode collapse, adaptive image normalisation, synthetic biomedical data, evaluation metrics and the downstream effect of generated data on classification. More broadly, I am interested in how generative AI can be evaluated and governed when it is integrated into organisational and expert workflows.

Workflow analytics and organisational AI

I am increasingly focused on AI-driven workflow analytics, data-driven decision support and the organisational consequences of AI adoption. This includes modelling how information moves through socio-technical systems, how AI changes decision processes, and how transparent AI can support digital transformation and the future of work.

Complex systems and information-theoretic modelling

My work on complex adaptive systems uses multimodal and psychophysiological data to study dynamic behaviour and emergent patterns. I am also interested in information-theoretic approaches to modelling and optimising data-intensive processes, particularly where conventional linear descriptions fail to capture system behaviour.

Biomedical, clinical and psychophysiological AI

Healthcare remains an important application domain. My work spans biomedical image analysis, physiological-signal analysis, ECG quality assessment, clinical decision support, affective computing, fairness and reproducible machine-learning pipelines. These applications provide demanding test cases for trustworthy AI because decisions must be traceable and data quality, bias and uncertainty have direct practical consequences.

Research directions for collaboration

Current topics of interest include:

• AI-driven workflow analytics for organisational and socio-technical systems
• Information-theoretic approaches to modelling and optimising data-intensive processes
• Complex-systems modelling and simulation of organisational decision environments
• Explainable, trustworthy and fair AI for workflow-based decision support
• Distributed and open-source AI frameworks for scalable analytics

Research affiliations

I am affiliated with Lero, ADAPT, and the Riomh research group at MTU. I also contribute to national and international research communities through conference organisation, programme committees, reviewing and postgraduate supervision.