Impact of built environments on human perception: A systematic review of physiological measures and machine learning

With rapid development in the field of artificial intelligence, an increasing number of studies are leveraging physiological and/or neurophysiological measures in conjunction with machine learning (ML) to explore human perception in built environments. This growing body of research has facilitated building design simulation and informed decision-making processes. However, a comprehensive review of this interdisciplinary field has not yet been undertaken. Thus, this study systematically reviews and critically evaluates the literature on the effects of built environments on human perception, specifically focusing on research that integrates physiological measures, subjective reports, and ML. The aim of this research is to develop holistic knowledge in this emerging domain while identifying research challenges and future directions. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, ten peer-reviewed journal articles are identified and analysed based on their research objectives, building and environmental contexts, subjective measurements, physiological measurements and ML algorithms. A comparative analysis synthesises findings across studies, demonstrating how ML models achieve high performance in analysing and predicting human perception in building design and simulation. Finally, this review identifies multiple research gaps and challenges, emphasising significant opportunities for future interdisciplinary exploration. This study contributes to the field by providing a structured synthesis of how machine-driven methodologies can enhance human-centred building design and decision-making in simulation. By offering a novel interdisciplinary perspective, it informs future research, design practices, and policy development in the built environment. • Research at the intersection of architecture, neuroscience, and AI was reviewed. • A systematic review and meta-analysis of ten research articles were conducted. • Physiological measures were synthesised with machine learning techniques. • Challenges for machine learning in building design and simulation were identified. • Research gaps and directions in machine-driven building design were highlighted.