As part of a project for a course I took in User Interface Design (INLS718) at the University of North Carolina at Chapel Hill, myself and two fellow students completed an evaluation project of the seemingly discontinued EarlySense Live Sleep Monitor Application. The EarlySense Live application is used in conjunction with a physical sensor device that is placed underneath your mattress. The device analyzes sleep cycles and monitors health while the user sleeps. It sends real-time updates of descriptive statistics about sleep stages, heart rate, breathing rate, stress level and movement to the application as you sleep. EarlySense was originally created to better support general care patients during their hospital stays but has since been used and marketed outside of hospitals to health-conscious adults, parents, and caretakers of senior citizens.
I’ve summarized the project on this page, but as of writing this our full project report is available online through the University of North Carolina at Chapel Hill’s Health Sciences Library website.
Preliminary issues were identified by trying out the application, reviewing customer product reviews on websites like Amazon, and conducting exploratory usability studies.
- Users have limited means to compare changes in their sleeping patterns or other health metrics over time in a way they can make sense of.
- Metrics, especially those related to sleep, are poorly explained and defined.
- The formula for scoring sleep is unclear in both what a “good” score means and how it is calculated and displayed graphically.
- Interacting with charts often causes scroll hijacking as they were not built in a mobile-friendly manner, despite being a mobile application.
- The homepage does not display any useful or relevant information.
Our goals for improving the system came from the issues identified to the left. While there were many issues with the system, the following were the key ones we wanted to improve:
- Make it easier and more intuitive for users to monitor the changes in their sleep quality over time.
- Allow users to find information about their vitals and sleep metrics easily and communicate that information in a manner that they can understand.
- Simplify navigation throughout the application so that users can find the information they need.
- Simplify graphical displays within the system to be easier to understand, as well as easier to interact with without disrupting other tasks.
Personas, Use Cases and Scenarios
After our initial review of the product and preliminary research, we developed personas, essential use cases, user scenarios and the tasks they would complete, and a hierarchical task analysis for each use case. The purpose of doing this was to identify which of the issues we’ve witnessed could have the most impact on how user-friendly the device is to their target audience.
We have developed 4 personas, 2 primary and 2 secondary personas, based on both the analysis
of users from EarlySense, as well as an analysis of common themes across customer reviews on
Amazon. The personas can be further divided into 2 types: those who want the device for their
own use, or those who want it to monitor others, where there is a primary and secondary persona
for each type. For more details on the above personas, you can read pages 7 through 10 in the project report.
We identified several use cases that the application had, from entering information in a diary to tracking sleep data. After conducting a hierarchical task analysis we ranked which use cases were the most essential for this application, which mostly involved tracking vitals, reviewing changes in sleep over time, and adjusting settings. Scenarios were developed for each essential use case based on the relevant personas.
Proposed Design Changes
Based on the tasks we identified and preliminary research, we worked on 6 key areas to improve usability on the application.
- Allow users to compare data from different days side-by-side.
- Provide explanations of metrics in the “sleep summary” tabs.
- Modify the sleep score formula and graphical display to be more clear.
- Add a way to expand charts and interact with them to avoid scroll hijacking.
- Update the homepage to have more relevant information.
- Display sleep information first on metrics sidebar.
While more detailed information can on all the above issues be found on pages 23 to 29 in the full report, I’ll summarize issue #5 here:
The existing homepage only displayed whether or not the sensor was connected, if the user was in bed, showed “live” tracking and had a “smart wake-up alarm” switch to enable that supposedly would sound within a time-frame when the user was in an appropriate phase of sleep. The live tracking aspect may have made some sense from the point of view of a parent tracking a child or a healthcare practitioner tracking a patient, but for somebody using this device themselves it did not seem useful at all as they would be asleep as it’s running. Even for a parent or healthcare practitioner, if there was a need to closely track vitals in a child or patient they would likely want to use more precise tools that have an alarm system for irregular metrics.
The redesigned homepage we came up with still included information about the device’s status and the user’s status as well as still allow users to turn the alarm off or on. However, instead of live tracking we opted to include information on the past few days for a quick accessible summary. This also addressed another issue of the application which was that after waking up you would have to click through a few pages to actually see your summarized results and scores. They should have instead been immediately available from the homepage.
Prototype and User Testing
The redesigned homepage screen is one example of a change we implemented in the modified application prototype. We developed other screens and tasks based on the 6 key issues we identified and then tested each of these tasks with 4 users following the testing script that can be found on pages 39-40 of the final report. We compared the results of usability testing to what we found in our preliminary research, and while there were many areas that improved, we identified some weaknesses:
- More information on sleep metrics and what they mean were helpful, but some additional information on steps to improve sleep could be even better. Also, it would help to include links to more detailed consumer-grade articles such as those found on WebMD.
- Our proposed new way of charting the sleep score, a waffle chart, was still confusing. While the totals for the scores made more sense, the chart displayed too much info and required users to count.
- The sidebar lacked clarity in how views could be swapped to see different metrics.
This project was the first time I did user testing that involved an at-home healthcare device, and there were unique challenges to evaluating an application that is supposed to be paired with an application. This was especially challenging for a device that is used during sleep, as we couldn’t feasibly test the device and application interaction by having participants use it while sleeping as it was a rented device and done as a class project with limited resources to involve participants. That being said, it was a great experience in trying to balance user-friendliness of the device with trying to provide people with accurate and useful medical information, which is often confusing. There was also the challenge of trying to represent complex data with several dimensions in a neat graph, which I don’t believe we fully succeeded in. That being said, the challenges of trying to visualize complex data and explain detailed information simply in this project have helped me improve my data visualization and UX writing skills.