Dr Emmanuel Boateng

My research is organised by the question each strand answers and who it serves, rather than by the methods used. Machine learning, deep learning, and conversational AI are the connecting craft across five lines of inquiry.

Safety and wellbeing on site. How work stays safe and workers stay well. Building on my doctoral research on co-worker safety interventions in construction, this line spans safety climate and safety behaviour, validated measurement instruments including a safety culture index for developing countries, and organisational approaches drawn from high reliability and resilience engineering. Its current centre is PsycheGuard, a conversational AI tool developed in collaboration with Queensland University of Technology that supports the mental health of construction workers, alongside continuing research on psychosocial risk in project-based work.

Delivering projects, cost and value. What makes projects buildable and worth building. This line covers tender pricing and cost estimation, value management, procurement capacity, and the competencies behind public-private partnerships and international construction joint ventures, much of it in developing-country contexts. It includes deep learning models that predict green building project cost and duration from early design data, developed with the Hong Kong Polytechnic University and implemented in industry.

Public voice and the classroom. Listening to what people actually say, at scale, to inform policy and teaching. Using text mining and sentiment analysis of large-scale public discourse, this work has examined public opinion on energy-efficiency retrofitting, decolonisation discourse in African higher education, and public interpretations of difficult histories and their implications for teaching. In the classroom, I lead work on intelligent learning assistants and feedforward agents that help students engage with assessment, particularly those who are introverted, multilingual, or supported by accessibility services.

Sustainability and the built environment. Modelling the built environment's footprint so it can be lowered. Published in journals including the Journal of Cleaner Production and Energy and Buildings, this line applies neural networks and other machine learning methods to carbon emission intensity, building-related emissions, energy demand forecasting, and the uptake of sustainable construction.

Foresight for health and environment. Forecasting risks before they arrive. This line projects the global prevalence of hypertension, diabetes and obesity to give health planners early sight of where the burden is heading, and models how persistent contaminants such as PFOS behave in soils to support environmental remediation, with work published in BMC Public Health and Environmental Science and Technology.

International collaboration is central across all five lines, with partners at the Hong Kong Polytechnic University, Queensland University of Technology, the University of Newcastle, and institutions in Ghana.