Device Prototype Project

We just posted a prototype sensor project summary on our QuickStartrm site. Using PiLR we built a first generation prototype of a mobile application that can help a manufacturer understand how users are interacting with their mobile or body worn sensor. We implemented 4 general functions: lifecycle usage tracking; reminder tracking; missing data attribution; and user reported device charging activity. For lifecycle usage our mobile app collected, processed and forwarded sensor data.  In this case we were monitoring a wrist worn multi-axis research grade accelerometer.   Warmer colors indicate more intensive usage.  No color indicates that no data was collected.  The attached chart shows what this visualization looks like.  Since users may purchase the device at any time, the days are relative days.  This is a simple way of visualizing user activity over time.  

We used a timer driven trigger notification to remind users to recharge their devices and tracked if they responded to the notification.  The trigger presented a survey to capture whether the user was successful in charging the device.  To track gaps in data, the mobile app monitored the device's output stream.  If the accelerometer readings were very low or null we triggered a survey asking the user why the device might not be transmitting data.  The user selected why the data was interrupted and we summarized their observations in the red chart.  This event/survey combination provides 2 functions.  It reminds the user that something they have done has interrupted the data (as an example, they are not wearing the device).  It also allows us (the designer) to build a ranked list of most likely to occur problems.  We can use this to alter the packaging and instructions so we get fewer periods with no data.  In addition, since we can process the data on the phone, you have the option of creating an intervention or reward on the phone.

All of the information is captured and forwarded to the PiLR platform in real-time. This gives the design team real-time access to device output.  This can be used to do population level analysis on usage patterns.  Since PiLR captures additional metadata, it is also possible to understand how good the connection is between the sensor and the phone.  These insights can be used to change device design, packaging, or in the creation of incentive programs to drive use engagement.  

PiLR adds value to the process by making the development and deployment of the mobile application very easy and fast. The total effort to create this prototype was less than 40 hours. Of this, 80% of the time was taken for the data analysis and visualizations. All the content and rules that create the mobile app were completed in about 8 hours. This yields a mobile app that is deployed in iTunes and the Google Play stores. All of the data and metadata are automatically generated. What's nice is that all of the streams have common notions of identity and time. The common infrastructure also makes it easier to combine data from different sources. As an example, it is easy to couple sensor data about activity with user observations about how they are feeling as both streams have common notions of time. And of course the system provides the type of security you would expect for these types of data.

Interested in learning more? Visit the QuickStartRM site.