Automatic Estimation of Fluid Volume Intake

Abstract

Sufficient water and fluid intake are necessary to sustain life and maintain good health. Monitoring of fluid intake of a person in a seamless manner is a challenging task, but it holds a great value in terms of accurate assessment of the drinking behavior of individuals. We report a method for monitoring fluid intake using wireless surface electromyography (sEMG). The first part of this work was conducted in a lab-controlled environment. We were able to estimate water volume intake with an accuracy of 94% using a linear regression model and Hilbert transform. The second part was carried out using a mobile sEMG device and we were able to estimate water intake using a Long Short-Term Memory algorithm (LSTM) with an accuracy of 92%. The relative immunity of sEMG as a sensing method and the reasonable accuracy we obtained using the mobile sEMG device suggest the feasibility of doing volume estimation of fluid intake in real-life setups.