Mentos
Background
More than half of Americans don't take medications as prescribed, and amongst this number, patients with type 2 diabetes have the lowest rate of medication adherence. Consequences of poor medication adherence for this specific group has the most serious outcomes-- hypoglycemia, falls, emergency hospitalizations.
With the sponsorship and help from Cognizant, our team explored various causes, identified key issues and narrowed our scope to resolve medication adherence for elderly patients with Type 2 Diabetes.
Solution
A virtual pet assistant that helps T2DM patients with disease management
A digital pet that tracks, reminds and encourages patients to adhere to medication regime to help elderly type 2 diabetics who lack social support and struggle with mental obstacles to adhere to their medications.
Simple, Error-free Setup
Organizing complex and frequently changing medication regimen with object-character recognition
Reminder and Confirmation for Complex Medication Regimens
Motivational reminders and confirmations to prevent mix-ups in medications.
Overview
Design Process
Our 3 month process consisted of biweekly presentations, presenting the work completed over time with our advisors and subject matter experts.
Research
Understand and present epidemiology, existing solutions and identified opportunities
Ideation
Brainstorm, decide and present 3 concepts based on design objectives
Refine
Expand and refine features of 1 concept, research existing features and usability test
Deliver
Iterate design based on feedback and present to stakeholders
Research
Why does this happen? Who is involved?
We conducted literature research and speaking with Cognizant's subject matter experts who were chief medical officers, HMO/PPO plan consultants and health technology experts to identify, and clarify main causes of medication non-adherence.
Though we did not have access to patients directly, there was a lot of primary literature already that were accessible, and we reached out to family, friends and friends of friends for further insights.
Through our literature studies-- we were able to break down causes of non-adherence in 5 dimension: mental, emotional, social, knowledge barriers and financial.
To understand when patients face such obstacles and to what frequency, we created a user journey map of a patient starting from when they are prescribed the medicine to the next doctors' visit. This helped us identify opportunities and stakeholders involved.
5 Causes of Non-Adherence
Poor self-perceived health status
Comorbidities
High cost of treatment
Fear of medication side-effects
Fear of injections
High treatment complexity
Understanding of disease
Forgetfulness
Lack of perceived benefit
Understanding of hypoglycemia
Unclear instructions from medical providers
Insufficient family support
Low trust in provider
ideation
Brainstorming ideas and refining concepts
We categorized our ideas by educational and emotional support to make sure they meet our design goals from our problem statement. Then, we categorized them by digital or physical solution to assess feasibility. We narrowed down to three ideas based on the the number of votes each team member assigned by how it could met our design goals.
Diabetes Community Chat App
A social platform specifically designed for diabetes patients that focuses on emotional support and having questions answered by medical professionals. Our research showed that peer support communities are often used by patients. Peer support communities, both online and in-person, can help fill in the gap of continuing support. Social communities that center around a disease state exist in online forums, Facebook groups, and disease-awareness organizations, which offer a sense of community and an outlet of support for the day-to-day challenges of managing diabetes.
RX Management Dashboard
Diabetic patients provides a daily overview of personal health status and reminds patients to take their medications. With more research, we also learned that behavioral changes occur with extensive knowledge of personal health. Mobile health platforms which display patients’ information empower patients with a sense of self-efficacy and motivate them to be proactive. Current existing solutions however, may be overwhelming and cause information overload.
Virtual Pet
We learned dog ownership has beneficial effects on physical activity and emotional well-being, both known to reduce the risk of comorbidities associated with Type 2 Diabetes, such as anxiety, depression, declinig physical activity and sociability. Pet concept may increase physical activity, providing companionship and add structure to the day. Emotional support could be a powerful motivator for patients to successfully take their medication and take an active interest in their personal health. The virtual pet application could remind them to take their medication, but also encourage them to engage in the physical activity that diabetics need.
user flow
Concept Decision
To decide the design direction of our final solution we first presented our solution our peers, advisor and professor. We saw many opportunities to provide features that would meet our outlined design objectives. Our key features to explore were: Physical Exercise, Medication Reminder, Learning Disease State, Mood Tracking
How to Encourage Daily Exercise
When we initially brainstormed the physical activity feature in the virtual pet application, we considered an AR experience that places a virtual pet on the camera screen where the user is, as it can provide an engaging experience for potential users. We believed that AR experiences could make users feel that the virtual pet is physical with the user. However, when we evaluated with our mentor, Marcia, the danger of the AR experience became a cause for our concern, we realized road injuries incurred by people playing video games on mobile phones while walking.
To avoid the potential risks that the application could present in an AR background, we decided to implement a bird-eye view map, although it is not as engaging as an AR experience. We thought a bird-eye view map would be effective because users are familiar with the map background and can easily track their location while walking. However, some of the feedback we received when we did initial concept was that the feature seemed unappealing and the experience was less immersive.
The final design we decided to come up with for the physical activity feature was to have a cartoon background that would allow virtual pets to be shown. Virtual pets will show their back as the user walks, as if the user felt like they were walking with their virtual pet. However, when the user stopped, we created a design that made the virtual pets look back at the users and indicate to them how their walking process is going.
For better engagement, we also decided to introduce a camera component where users can take pictures of their virtual pets using the camera feature on their phone. Virtual pets will show up on the user’s camera screen. In this way, users would feel engaged and immersed in the experience. Based on research and discussions with mentors and professors, we realized that it would be a good idea to have these features, as users wouldn't have to pay attention to the app while walking, but they can easily choose to interact with their virtual pet at any stage of physical activity.
Complex Setup Experience for Frequently Changing Regimen
As the primary goal of our application is for medication adherence, we wanted to make sure our application addressed the existing issues that were present in existing medication reminder systems, while simultaneously thinking about how to increase adherence, not from a reminding standpoint. In our initial research, we discovered the following pain points in digital pill reminding systems: a complex setup experience and lack of motivation.
Leverage OCR for Easier Set-Up
The idea we moved forward with was to simplify the medication setup process by leveraging optical character recognition (OCR). Instead of having patients type in their prescription information that can change frequently, patients could simply take a picture of their medication label. The application on the back end can use OCR to correctly
recognize the directions for medication and set up reminders as needed. We were able to test this by using Google’s Cloud Vision API. By adding a link to this example label image from CVS, the API was able to return a string of text that accurately matched what existed in the label, which we thought would be a promising solution to reduce the setup burden for patients.
Connecting with EMR/Pharmacy Software
Another consideration we made as to simplify the experience by connecting with doctors’ EMR software and pharmacy software. This would require no effort on the patients’ end to add their medication information. Additionally, the information that the application collects from the doctor and pharmacy software would always be up to date.
refine based on testing
Test Methods
Cognitive Walk-Through
We decided to create high-fidelity wireframes from our original mid-fidelity wireframes before usability testing, as our key tasks were scanning the pill and vial, which would involve a high fidelity image from a camera, and an AR experience that would also incorporate a real photo. Instead, we completed a cognitive walk-through with our professor and advisor to validate the sequence of screens in our mid-fidelity wireframes. We then incorporated feedback, created high fidelity wireframes, and then planned our usability testing.
Usability Testing
Our primary testing goals were to make sure the app user flow worked as intended, confirm that it was easy to use, and simple that users were focused on the immediate task they had to complete on the application. We tested 5 key flows which were logging in, medication setup, taking medication, checking mood, and going for a walk. We used 5 points Likert scale to gather quantitative data on users’ ease of use for each user flow. We also gathered qualitative data by asking open-ended questions to learn what part was memorable, noticeable, and delightful. We recruited 5 testers from family and family friends that met our target population demographics. 4/5 testers had type 2 diabetes, and 1/5 of testers were not diabetic but took chronic medications.
deliver
Final Prototype
Final Thoughts
Humility - Designing in the healthcare space is complex
Healthcare design is very complex, and shifting people's attitudes towards it is not easy. To reduce the mental load and making medication adherence feel like a chore, we explored how we can shift that mindset to a more fun approach.
Design for all vs. Feature bloat
As we narrowed our target group to T2DM medications for its potential impact, we also learned the special difficulties T2DM patients face with medication adherence. T2DM patients have complex and frequently changing medication regimens that are hard to keep track of, especially if patients have hospital visits, changing physicians, and with age- less cognitive ability and social support.
A learning point from this project was - to effectively prioritize features. As a conceptual project, it was very tempting to try to solve for every cause which led to a
feature-bloated product. I would also explore other IoT solutions that could make the process easier. For example, maybe medication confirmation can be done through a Nest camera, or a tablet installed in a wall so users can freely use both hands. Maybe the pet can be more life-like if it was projected on a TV screen.
Technical v. Emotional solutions
My biggest takeaway from this project was that some solutions warrant a more emotional approach. Discovering that patients use facebook forums and reddit groups for support in their diabetes journey was a huge learning point. That medication adherence, or patient attitude towards better health may not be an issue of memory, rather a social and emotional one.
Special thanks - Marcia (Healthcare IT SME), Wendell - (Advisor), Cognizant (Sponsor), Linda Design, Yuyang Design Illustration
View Process Book →