Cacophony Mapper

Negative Sound Experiences of Nurses in the Emergency Room: Noise Fatigue

Noise fatigue is a medical condition observed in nurses who have been exposed to prolonged noise. It has been reported that healthcare professionals who are continuously exposed to negative noise fatigue in the emergency room experience both physical and mental symptoms. In addition to risk factors such as decreased job satisfaction and hearing loss for healthcare professionals, alarms in the emergency room are closely related to patients' health conditions, potentially causing dangers in the ER. In this context, I have designed a system to analyze and improve the sound experiences of healthcare professionals in the intensive care unit of Erasmus University Medical Center in Rotterdam, for which they are a client. If I wanted to be a designer

System Architecture of the Cacophony Mapper

The Cacophony Mapper device is equipped with various sensors to collect multimodal data closely related to the sound experiences of healthcare professionals. The Cacophony Mapper gathers heart rate data from the outside of the nurse's arm through a wearable device, which is closely related to the stress levels of healthcare professionals. Additionally, it simultaneously collects sound data through a microphone component attached to the nurse's chest pocket. Furthermore, the system is designed to enable real-time emotional reporting by nurses via a mobile application, and it collects and maps indoor location information in real-time using beacon data attached to the walls of the emergency room.

Data from various layers was stored in the Firebase database in real-time and utilized for machine learning model training and data analysis. I divided the Erasmus University Hospital emergency room into three sections to identify the locations of nurses using beacon data. Emotion reporting aimed to understand nurses' emotions toward unpleasant and stimulating sounds located in the first quadrant of the Circumplex Model of Affect. It allowed real-time reporting of five emotions — irritation, anger, annoyance, tension, and discomfort — with the click of a button. First, sound data collected in the emergency room was processed through a fast Fourier transform filter, which enabled the training of a classification model that could distinguish between four categories of sounds: mechanical noise, alarm sounds, conversation noise, and background noise. This process allowed for real-time identification of which category incoming sound data belonged to. Additionally, heart rate data, known to be closely related to the stress and tension levels of medical staff, was also used in data mapping. By visualizing multimodal data from various layers in real-time, we enabled a more intuitive analysis of the sound experiences of the medical staff and provided solutions for negative sound experiences.