Improving Occupant’s sleep quality with the help of OURA ring and data from Smart Buildings

Abstract: Well-being is associated with comfort and health, and it represents wellness and quality of life. Sleep quality is an important index when evaluating a person’s well-being. KTH Live-in-lab performs Human-building interaction studies to explore the growing potential of how built environments, measured by Schneider Electric (SE), can influence humans and their well-being in their everyday lives. This thesis works as an explorative study of using the OURA ring to evaluate sleep quality for tenants living in KTH LiL. Specifically, this project aims to assess the quality of the data collected from the ring and SE sensors by using Total Data Quality Management (TDQM) and propose a Multilayer perceptron (MLP) model for predicting sleep scores. Results first showed that the OURA ring is an appropriate tool for evaluating sleep quality. Its data passed 11 TDQM’s dimensions, including accuracy, objectivity, relevancy, interpretability and understandability. Second, the OURA was able to capture the relationship between sleep quality and building’s temperature and humidity through its sleep scores. Results showed that higher sleep scores situated more around the suggested ideal ranges of temperature and humidity. However, some low sleep scores were also situated around these ideal ranges which suggests that an additional study needs to be conducted. Such a study would take in tenants’ feedback in order to distinguish sleep scores heavily affected by psychological and/or other factors rather than built environments. Third, we were able to create an MLP model to predict sleep scores based on temperature and humidity values as well as user-related information, like activity rate and total burn. The model had validation and training losses converging at 1.90-2.50. Those low loss rates suggest that the building's temperature and humidity along with information about tenants from the ring can be used to improve the sleep scores. This model can be extended into a recommendation model where buildings’ operators and tenants can benefit from. Buildings’ operators would get information and recommendations on how to properly administer their buildings to achieve higher well-being for their tenants. Also, tenants would get recommendations on how to increase their sleep scores and, ultimately, their sleep qualities and well-being.