Towards a monitoring and emergency alarm system activated by the barking of assistant dogs

The quality of life of people who are susceptible to presenting a sudden medical emergency is reduced by constantly suffering from a mental state of insecurity and concern, both on the part of patients and their caregivers. This is the case for older adults, people who suffer epileptic seizures, and people with emotional crises, to name a few. Assistance dogs trained to live with such people play a significant role, as they can identify emergencies and carry out specific actions to help the patient. It has a positive impact on people's sense of security and self-esteem. In this article, we present the design, implementation, and validation of an environmental audio monitoring system that recognizes barking and detects the marking bark pattern of trained dogs to notify emergencies. When the device detects the barking pattern, it sends out an alarm signal that can be received on a caregiver's phone. We evaluated the automatic bark detection performance of the audio classifier using the F1-score accuracy metric obtaining an average of 94% under ideal conditions, and 98% in a house simulation environment using background noise. Regarding the evaluation of the alarm algorithm, the system obtained 85% without background noise and 82% with background noise. This project intends to create the technological foundations for a system that improves the quality of life for patients and caregivers by enhancing the capabilities and maximizing the talents of assistance dogs taking advantage of the impressive current advances of information and communication technologies.