Finding the Proper Mental Stress Model Depending on Context using Edge Devices and Machine Learning

Aim of the paper is to demonstrate how finding the proper mental stress model and its dependencies on context using Machine Learning technologies resident on edge devices to allow the user and remote medical staff to continuously monitor and control the user stress status. To this aim, the paper first discusses the method used to measure the mental stress inspired by tools available on the market, secondly it illustrates how the sensed bio-data should be preprocessed on an edge device to support the first control actions. Finally it shows how such data may used to derive a stress model of the user using a machine learning algorithm on edge devices and/or computing server. An example illustrates the proposed methodology, how this model can be tuned depending on context using the data collected by the wearable monitoring device, and how the entire system can be implemented on few interconnected micro-boards. A case study demonstrates how deriving the mental stress model of a subject depending on contexts also by using the stress mental model derived from a community of people.