BeWell: A Smartphone Application to Monitor, Model and Promote Wellbeing

Nicholas D. Lane, Mashfiqui Mohammod, Mu Lin, Xiaochao Yang, Hong Lu, Shahid Ali, Afsaneh Doryab, Ethan M. Berke, Tanzeem Choudhury, Andrew T. Campbell2011 · DOI

Abstract

A key challenge for mobile health is to develop new technology that can assist individuals in maintaining a healthy lifestyle by keeping track of their everyday behaviors. Smartphones embedded with a wide variety of sensors are enabling a new generation of personal health applications that can automatically monitor and model user behavior.

Introduction

In the early 2010s, the "Quantified Self" movement highlighted a significant barrier in mobile health (mHealth): friction. Most health applications at the time required manual data entry or expensive external wearables, leading to high abandonment rates. Furthermore, health was often viewed through a narrow lens—focusing solely on physical activity while ignoring the critical roles of sleep and social connectivity in mental and physical longevity.

BeWell addresses these challenges by providing a system that monitors three pillars of health—sleep, physical activity, and social interaction—continuously and passively using only the sensors found in a standard smartphone (accelerometer, microphone, and GPS).

System Architecture and Methods

The BeWell system is designed for "near zero user effort," utilizing a split architecture between the mobile device and cloud-based services to balance real-time feedback with complex data modeling.

The BeWell system architecture, illustrating the Sensing Daemon on the smartphone and the supporting Cloud Infrastructure for data storage and modeling.

Passive Sensing Dimensions

  • Activity Tracking: The system uses the accelerometer to classify behaviors such as walking, running, or remaining stationary via a Naive Bayes classifier. These activities are mapped to Metabolic Equivalent of Task (MET) values to estimate physical exertion.
  • Social Interaction: To protect privacy, the microphone detects "voicing" segments without recording raw audio. The duration of detected human speech serves as a proxy for social engagement.
  • Sleep Modeling: BeWell infers sleep duration by monitoring phone charging events, ambient silence, and periods of phone inactivity. The model uses a Gaussian function to penalize both insufficient and excessive sleep, aligning with medical consensus.

Ambient Feedback and User Interface

A core innovation of BeWell is its use of "glanceable" feedback. Rather than overwhelming users with complex charts, the application features an aquatic-themed ambient display as the phone's wallpaper. This allows for the passive consumption of health data, designed to influence behavior subconsciously.

The BeWell ambient display showing the evolution of the aquatic ecosystem from low wellbeing scores (left) to high wellbeing scores (right).

In this ecosystem, different animals represent health dimensions: a turtle represents sleep (hiding in its shell when the user is sleep-deprived), a clownfish represents physical activity (moving faster with more exercise), and a school of fish represents social interaction (growing in size as socialization increases).

Evaluation and Results

The system was evaluated through a week-long "in-the-wild" deployment with five participants using Android Nexus One smartphones. Key findings include:

  • High Accuracy: The system identified physical activities with accuracies ranging from 85.3% (voicing) to 98.1% (running).
  • Sleep Estimation: The proxy model achieved a Mean Absolute Error (MAE) of approximately 1.5 hours, which the authors deemed adequate for general wellbeing monitoring.
  • Battery Performance: Despite continuous sensing, the application maintained a battery life of over 15 hours, ensuring it could last a full day between charges on 2011-era hardware.

Discussion and Future Work

While BeWell proved the feasibility of holistic, passive health monitoring, it faced limitations such as the "TV Problem," where ambient noise could be mistaken for social interaction. Future iterations aim to include more robust conversation detection and additional health dimensions like diet and stress. By moving away from burdensome manual entry, BeWell represents an early pioneer in making health tracking a seamless part of daily life.

Summary updated

Abstract

A key challenge for mobile health is to develop new technology that can assist individuals in maintaining a healthy lifestyle by keeping track of their everyday behaviors. Smartphones embedded with a wide variety of sensors are enabling a new generation of personal health applications that can activ
BibTeX
@misc{lane2011bewell,
  author = {Nicholas D. Lane and Mashfiqui Mohammod and Mu Lin and Xiaochao Yang and Hong Lu and Shahid Ali and Afsaneh Doryab and Ethan M. Berke and Tanzeem Choudhury and Andrew T. Campbell},
  title = {{BeWell: A Smartphone Application to Monitor, Model and Promote Wellbeing}},
  year = {2011},
  doi = {10.4108/icst.pervasivehealth.2011.246161},
}
Figure