Research Projects

The well-being and health of any infant are of paramount importance for new parents. Unfortunately, due to unavoidable reasons, many infants are born prematurely or face critical illness. This necessitates their stay and proper care to be taken in the neonatal intensive care unit (NICU). However, the usage of health monitoring devices in the NICU exposes these babies to high dB noise exposure, which is extremely hazardous for their health, often at levels of 50-60 dB, which is quite hazardous. This has been observed to have an adverse impact on sleep patterns, auditory loss, critical physiological issues such as variation in blood pressure, heart rate, oxygen concentration, sugar level, etc. Further, this may also lead to long-term acoustic-related issues or profound deafness in the infants. This poses to be a serious issue towards physical, physiological, and neurological development of the infants in the long term, affecting the development of their vital organs and longevity. There are multiple and varied noise sources in NICU, for instance, from medical equipment such as incubators, ventilators, monitoring devices, IV infusion pumps, etc. The alarms and beeps from monitoring devices such as oximeters, cardiac monitors, etc, send any warning or alert signal when encountering an aberration or deviation. In addition, the movement of staff and caretakers also add up to the inclusion of noise inside the NICU. Henceforth, in the NICU, the usage of multiple monitoring devices along with various incubators calls for effective noise alleviation at various spatial regions, specifically at the location of each and every incubator. Therefore, this project proposes to mitigate the impact of NICU noises employing a distributed active noise control technique in conjunction with a deep learning framework. Henceforth, in lieu of this, the project proposes to utilize the notion of distributed active noise control, visualizing each incubator as a node of the entire distributed set-up. The notion is to leverage the benefits of utilizing each of the incubator units as a wireless acoustic sensor node (WASN) since it is accompanied with a speaker, error microphone, and active noise cancellation (ANC) unit. Each node facilitates information flow among the adjacent nodes and cooperatively learn to reach to a global solution. Deep learning is incorporated in ANC with the purpose of adapting to complex, uncertain noise characteristics, avoiding the need for secondary path estimation. This further enhances the real-time adaptability, modelling the non-linear and varied noise patterns with faster convergence and higher precision. The project aims to enhance the quality of services provided at the NICU by maintaining the noise level of the essential healthcare monitoring devices.