A Measurement Approach for the Assessment of Indoor Multidomain Comfort: The MULTICLIMACT Experience

The evaluation of indoor multidomain comfort is pivotal from a dual point of view: i) knowing the actual occupants' status enables taking actions for improving it; ii) personal feedback (both objective - inferred through sensors - and subjective - collected through surveys) can be exploited for a better management of the building. This work presents the results obtained from analyzing data collected in an experimental test campaign conducted on 15 healthy voluntary participants. Participants were not assigned a predefined task sequence but were observed during their normal office routine; this included activities such as typing on a keyboard, working at a computer, writing by hand, and using a mobile phone, to capture physiological and environmental signals in realistic working conditions. Data recording was performed in a room of 42 m2 while wearing a smartband (Emotibit) collecting diverse physiological signals (e.g., photoplethysmogram and electrodermal activity); furthermore, a multiparametric environmental sensor was positioned in the room. A survey on comfort (thermal, acoustic, and air quality) and related individual perception and affect (positive and negative) was administered to the participants at regular intervals. A BIM-based platform was used to perform the tests, administer the surveys, and collect all the signals from all the sensors (even when multiple subjects were present in the room). All the collected data were post-processed; after specific cleaning operations, relevant features were extracted from both physiological and environmental signals and used to train Machine Learning modEIS to predict the subject's multidomain comfort sensation; also, synthetic indices combining multiple spheres of comfort were defined to depict the global subject's comfort status. The proposed measurement approach can be applied to regularly monitor indoor multidomain comfort, enabling timely interventions in case needed, also providing early warnings and suggesting actions for improvement.