Indoor temperature prediction methods for urban overheating: A systematic review

City-scale indoor overheating assessment is limited by the lack of methods that are simultaneously scalable, equitable, and applicable across diverse residential building stocks. This review evaluates existing indoor temperature prediction methods for this purpose, particularly given that outdoor temperature proxies explain less than 40% of indoor temperature variability and may therefore misrepresent population exposure. Using a PRISMA-guided review, 46 peer-reviewed studies published between 2015 and 2025 were analysed from an initial 492 records and classified into five methodological categories: physics-based simulations (32.6%), statistical methods (19.6%), machine learning (17.4%), deep learning (8.7%), and hybrid approaches (21.7%). These categories were evaluated against five criteria relevant to city-scale implementation: data accessibility, computational efficiency, scalability, accuracy, and transferability. The review found that no single methodological category consistently satisfied all criteria. Physics-based models were constrained by extensive input requirements and computational burden, statistical methods by historical data, machine learning and deep learning by limited transferability and sensor requirements, and hybrid methods by data availability and cross-context validation. A persistent equity gap was identified because methods requiring intensive monitoring infrastructure were least applicable to vulnerable housing. In response, this review proposes a hybrid archetype framework that integrates computer vision-based building characterisation, urban datasets, and planning records to capture fine-grained morphological and thermal diversity across large residential stocks, enabling archetype construction from representative rather than dwelling-level monitoring data. By coupling data-rich characterisation with physics-based thermal simulation, the framework aims to reduce the oversimplification of traditional archetype approaches and extend coverage to vulnerable housing at city scale.