Cox proportional hazards model used for predictive analysis of the energy consumption of healthcare buildings

Abstract

The energy consumption of healthcare buildings is very high due to their 24/7 functioning and the demand of electro-medical equipment. This excessive energy consumption cannot be attributed to a single factor. The objective of this research was to apply Cox proportional hazards model to predict the probability of energy overconsumption in healthcare buildings for different functional variables. A reference energy consumption index was established from a retrospective analysis of the monthly consumption of 64 healthcare buildings during the period 2015–2019. Functional parameters (construction, facilities, demographics, and climate) were selected as candidates for Cox proportional hazards model. The study found that the variables related to facilities and demographics significantly influence the semi-parametric model of energy consumption. Their influence was quantified, and the validity of the proposed model was verified graphically. Having more than 10,000 users was found to result in a 124% greater probability of exceeding the reference energy consumption, 97.4% greater with an installed power above 60 kW, and 94.6% greater if the town in which the healthcare building is located has more than 5000 inhabitants, and 69.4% greater if a heat pump is not used for air-conditioning. The Cox proportional hazards model was shown to be an advanced tool useful for quantifying the influence of various functional variables on the excess energy consumption of healthcare buildings. The results of the research generate objective information to establish, on the one hand, criteria for designing and renovating healthcare buildings and, on the other hand, care planning strategies.