Clustering acoustical measurement data in pediatric hospital units

The previous hospital acoustic literature has highlighted some important considerations and various complexities regarding objective noise measurements. However, extensive use of conventional acoustical metrics such as logarithmically averaged equivalent sound pressure levels (L_eq) do not sufficiently describe hospital acoustical environments and often lack considerations of the room-based activity status that can significantly influence the soundscape. The goal of this study was to explore utilizing statistical clustering techniques in healthcare settings with a particular aim of identifying room-activity conditions. The acoustic measurements were conducted in the patient rooms of two pediatric hospital units and subsequently classified based on two room-activity conditions—active and non-active conditions—by applying statistical clustering analyses with standard k-means and fuzzy c-means algorithms. The results of this study demonstrate the most probable noise levels and degree of associations of the measured noise levels for the two room-activity conditions. The results were further validated in terms of the clustered levels, the number of conditions, and parameter dependency. The clustering approach allows for a more thorough soundscape characterization than single-number level descriptors alone by providing a method of identifying and describing the noise levels associated with typical, intrinsic activity conditions experienced by occupants.