home portfolio teaching publications reading cooking
mobile health and wellness
projects all fall into the domain of mobile health and wellness, but
were not a part of a coherent research program. Wellbeing has always
been a side interest of mine and when possible, I have tried to create
interesting systems to promote awareness of one's body and wellbeing
that can help to motivate positive behavior changes.
CONCEPT DEVELOPMENT (2010-2011)
||People now have access to many sources of information about their health and wellbeing. From web-connected scales to pedometers, GPS workout units, food intake data, calendar appointments, and location information, people are being overwhelmed by the data that is automatically collected and can relate to their wellbeing.
Do I lose weight when I have busy days? Do I walk more when I work out of the office in the city? Do I sleep better on nights after highly-scheduled days? These are all questions people might have about their wellbeing, but existing tools did not provide the integration between data streams to identify these interesting events or allow for user queries into mashups of wellbeing data.
We built the Health Mashups system to provide a tool that can identify
stastically significant correlations between weight, sleep, daily step
count, calendar free/busy data, and location and surface these
significant items for users through a widget on their mobile phones.
They could then dig deeper to understand the correlations (e.g. You wake up fewer times during the night on nights when you exercised during the day) or
deviations (e.g. Yesterday you walked 2000 steps
fewer than average for a Tuesday).
We ran a two site field trial of this system for two months in the daily lives of ten participants in Chicago and Stockholm. We learned how to make the system more proactive to encourage engagement as well as how the data let people reflect on their wellbeing across aspects of their lives which no participant was able to do during the first month with the sensors and accompanying websites, but without the mashups system.
Personal Health Mashups: Mining significant observations from wellbeing data and context. CHI 2012 workshop on Personal Informatics in Practice: Improving Quality of Life Through Data. May, 2012.
Mobile Health Mashups: Making sense of multiple streams of wellbeing and contextual data for presentation on a mobile device. Konrad Tollmar, Frank Bentley, and Cristobal Viedma. Pervasive Health 2012. May 2012
Health Mashups: Helping People Find Long-Term Trends Between Wellbeing and Activities in Their Lives. Frank Bentley. Quantified Self Blog. April 17, 2012.
temperature sensing glove
CONCEPT DEVELOPMENT (2010)
||In 2010 around the time I started hacking with an Arduino in my spare time, a friend was having problem with arthritis in her hand. She wanted to know what factors led to it becomming inflamed and I had just the idea to do that.
I built a glove that has temperature sensors sewn into the fingers and sends readings via Bluetooth to an Android phone using the Amarino toolkit. The mobile phone stores the timestamped finger joint temperatures and also uses the Google Weather API to access the current temperature and humidity based on the phone's location. All of this data is logged to a CSV file on the phone for later analysis on a computer to find correlations between contextual factors and joint temperature.
In the end, this system demonstrated the ease of gathering health data from wearables and integrating them with data captured on a mobile phone for later analysis.
CONCEPT DEVELOPMENT (2012)
2012 Health and Wellness Innovation Hackathon at the MIT Media Lab, we created a diabetes-monitoring application called GluBalloon. We used the metaphor of a hot air balloon to represent a patient's blood-glucose level. The balloon would rise into the clouds if it was too high, or head towards a mountain range if it was too low. Sandbags off the side of the balloon represented factors that could lower the blood glucose levels: activity (step count), and insulin. In the balloon were photos of food that the user ate which cause glucose levels to rise.
The system automatically captured ground-truth step count activity on a MOTOACTV Android watch, glucose from a connected glucometer, insulin from a bluetooth-augmented insulin pen, and photos of food from a simple Android application we created. An simple interface on the MOTOACTV watch (top) and a more detailed view on a tablet (bottom) allowed users to reflect on the data and see their readings from anywhere.
We validated the design of the system with clinicians from the Joslin Diabetes Center as well as with individuals suffering from the condition. All saw the value in automated logging and the benefits to patients when they had an accurate log to reflect on and understand how all factors play together in their lives. The hope is that a system like this would help patients keep their glucose levels in the proper range as well as be more active so that they need less injected insulin.
GluBalloon: An Unobtrusive and Educational Way to Better Understand One's Diabetes Angelika Dohr, Jeff Engler, Frank Bentley, Richard Whalley. Poster/Demo/Video at Ubicomp 2012, Pittsburgh, PA USA. September 2012.
CONCEPT DEVELOPMENT (2011)
2011 Health and Wellness Innovation Hackathon at the MIT Media Lab, I
worked with researchers from KTH University and Humana to create the
StepWatch system, an Android-based watch that kept track of
minute-by-minute stepcounts. By extending the hour hand forward or
backward, the system showed at a glance how far ahead or behind the
average step count the wearer was for the current time of day.
We futher extended the Indivo health record system in order to support storing and querying real-time sensor data such as is needed for this type of application.
application allowed doctors to get time-critical patent information
on their two-way pagers or fax machines. This was part of a late-stage
accelerator that operated as a startup company inside of
Motorola from 1998-2000.
Through my internships, I designed and built the pager client which
in- volved designing the user interface as well as implementing an
encryption scheme ans modifying our server component to support
In a later summer, I worked on creating requirements
for version 2 of the system based on user feedback from a field trial
and my knowledge of the system from being one of the original four people on
As a result of the system, drug order cycle time was reduced from
27 hours to 4 hours after a lab test was completed in real hospital settings.