We describe Hydroprompt, a prototype for sensing and motivating healthy water intake in work environments. In a 3-week field deployment of Hydroprompt, we evaluate the effectiveness of three approaches to behavior change: historical information enabling users to compare their water intake lev- els across different times of day and days of week, implicit feedback providing subtle cues to users on the current hydration levels, and explicit prompting at- tempting to remind participants when hydration falls below acceptable levels or when substantial amount of time has elapsed since the last sip.
💡 Deep Analysis
📄 Full Content
Water is essential to our everyday functioning. Over 50% of the human body consists of water and an average sized person should drink at least 2-3 liters of water per day in order to remain hydrated [1]. However, recent studies have revealed that both adults and children often fail to maintain appropriate levels of hydration throughout the day [3,4].
One of the most effective triggers for water intake is thirst, the sensation of needing to drink [2]. Yet this signal is triggered when there is already a water deficit [2], thus, technology may play a role in establishing healthy water intake habits, and a number of attempts have been made towards this direction. For instance, playful bottle [5] senses water consumption through a mobile phone attached to an everyday drinking mug and attempts to influence users’ habits through implicit feedback. Hydracoach (www.hydracoach.com
) is a commercial bottle that monitors water consumption and provides historical information such as the average water consumption per hour and the amount of time elapsed since the last sip.
Yet, despite some recent interest on the topic, the domain currently lacks an understanding of the effectiveness of different techniques in motivating healthy water intake, with recent work in the broader area of behavior change tools raising concerns over the long-term effectiveness of existing approaches (see [6,7]).
In the current paper we describe the design and evaluation of Hydroprompt, a prototype for sensing and motivating healthy water intake in work environments. In a 3week field deployment of Hydroprompt, we evaluate the effectiveness of three approaches to behavior change: historical information enabling users to compare their water intake levels across different times of day and days of week, implicit feedback providing subtle cues to users on the current hydration levels, and explicit prompting attempting to remind participants when hydration falls below acceptable levels or when substantial amount of time has elapsed since the last sip. In addition, we attempt to measure the impact of subliminal conditioning [8] (through presenting positive phrases such as “water is good”) on the effectiveness of prompting.
Hydroprompt employs a load sensor from a regular kitchen scale for sensing water intake (see figure 1). Signal from the load sensor of the scale is led to an operational amplifier and then read by the Arduino platform. This communicates with a C# application that serves as the front-end of Hydroprompt. For successful sensing, the user has to continuously place her bottle or mug on top of the scale. Following the bottle’s removal from the scale, the platform senses any difference in weight and records the amount of water intake, if any. The proposed approach offers a simple, cost-effective and reliable sensing platform for water intake applications in stationary settings, such as office environments. Hydroprompt records the date, time and quantity of each sip and attempts to motivate healthy water intake in three ways: prompting, historical information and implicit feedback.
Prompting. Hydroprompt alerts the user at regular time intervals using a notification at the bottom right part of the user’s screen (see figure 1c). The notification consists of a glanceable visualization of the user’s hydration level and a short sentence relying on the principle of subliminal conditioning [8]. The idea of subliminal conditioning suggests that priming behavioral concepts (e.g., drinking water) motivates individuals outside conscious awareness, especially when primes match a current need (e.g., fluid deprivation) [9]. A collection of 10 short sentences were created (such as “water is good”, “Drinking water helps you feel more energetic”, “Drinking water can make you more productive”) and were interchanged during different notifications. The frequency of prompting depends on two variables: a) the hydration level of the individual at any given moment, and b) users’ stated preference about prompting frequency. Three levels of hydration were defined by the threshold values of 20% and 80%.
Historical Information. Hydroprompt presents -using graphs -the amount of water consumed over the course of a week, a day or recent individual sips. In addition, Hydroprompt presents the hydration level of the user at a given moment in time (see figure 3) as well as the extent of goal completion for the day (e.g., 760 of 1140 mL).
Implicit feedback. Hydroprompt attempts to provide implicit feedback on individuals’ hydration state through altering the wallpaper of their computer environments.
The system uses 5 different wallpapers, varying in hydration, to represent five levels of hydration (i.e., above 80%, 60%, 40%, 20%, 0%, see figure 3).
We conducted a three-week-long deployment with 6 participants (5 female, all office workers), using an ABA study design, with the goal of assessing the impact of Hydroprompt on individuals’ water intake behaviors. The first