HydrateME: A personalised hydration management tool based on machine learning approach

Hydration monitoring is a critical challenge in healthcare, impacting various groups such as athletes, soldiers in extreme environments, individuals with adipsia (lack of thirst sensation), and non-verbal elderly individuals. This project seeks to address this issue by leveraging wearable sensors and machine learning techniques to create dehydration detection models.

The project will use existing data collected from various sensors, including accelerometers, magnetometers, gyroscopes, galvanic skin response sensors, photo plethysmography sensors, temperature sensors, and barometric pressure sensors. Additionally, personal bio data will be incorporated to predict a person's last drinking time. Prediction beyond a predefined threshold can be used to issue an alert to individuals and their carers.

The research will explore a range of machine learning algorithms, considering both generic (one-size-fits-all) and personalized approaches. By comparing the performance of these models, the project aims to select the most suitable algorithms for on-device deployment optimization. Ultimately, this project addresses the pressing need for proactive hydration monitoring, benefiting a diverse set of individuals in various contexts. It combines wearable sensor technology with advanced data analysis to enhance healthcare services and improve the wellbeing of those susceptible to dehydration.

Funder: Research Investment Funds
Award: £20,000