Blume - Democratizing Therapy

Using AI to deliver a mental health support app that adapts with the users ever changing needs.

February 2024 - March 2024

Mental health concerns are prevalent globally, affecting an increasing amount of people. However, access to qualified mental health professionals remains limited due to factors like cost, availability, and geographic location. This gap highlights the need for innovative solutions that offer support to a wider population. AI-powered therapy shows potential to address this need by providing accessible, user-friendly, and personalized support.

Overview

The limited accessibility of qualified mental health professionals creates a significant barrier for individuals seeking support for their mental well-being. This lack of access can exacerbate existing mental health challenges and reduce overall well-being. My challenge was to leverage AI technology to create a user-centered solution to fill this gap for people seeking support for mental health and well-being. For this project I had 80 hours to work through the design process from start to finish.

The Challenge

An AI-powered app can offer these individuals assistance in keeping themselves mentally healthy. It can provide users with exercises and personalized plans to help improve their symptoms. My goal was to create an end-to-end MVP (minimum viable product) app to address this issue and help those in need.

The Goal

User Research, UX/UI Design, Visual Design, Branding Information Architecture, Prototyping, Usability Testing.

My Role

Tools Used

Figma, Maze, Notion, Illustrator

Team

Solo project

Process

1. Research

  • Competitive Analysis

  • User Interviews

2. Define

  • Information Architecture

3. Design

  • Branding

  • Wireframing

4. Test

  • Usability Testing

Research

Competitive Analysis

From an initial overview of 17 existing apps I narrowed this analysis down to 4, Earkick, Wysa, Elomia, and Headspace.

This revealed strengths and weaknesses, gaps, and common features amongst competitors. What stood out was the abundance of companies all delivering a similar solution, however, just seeming to fall short and hold dominance over others.

Key takeaways:

  • AI-assisted mental health apps are a valuable tool and have been proven effective.

  • These apps use various techniques like CBT, mindfulness exercises, and mood tracking to address anxiety, depression, and panic attacks.

  • There's a lot of competition in the market with most apps taking a similar approach, leaving an opportunity to stand out.

  • Based on the feature breakdown it seems “mood check-ins” and “exercises” are table stakes. Opportunities might lie in some form of a personalized coach and a unique approach to the check-in process.

I conducted user interviews with three participants within my target users. These proved very valuable in understanding users metal models, their needs, and pain points.

Interviews were outlined based on my research questions:

  • What are the main challenges individuals face when seeking mental health support?

  • How comfortable are potential users with utilizing an AI-powered therapy app?

  • What features and functionalities do users consider most important in a mental health app?

User Interviews

With a mound of quotes and observations from these interviews I created an affinity map to assist in extracting important themes and insights. These insights are what initiated the direction of the designs to follow.

— Affinity Map

Key Research Insights

Existing mental health apps often lack personalized support and depth.

Concerns regarding AI-powered mental health apps revolve around accuracy, user engagement, and skepticism towards chatbots.

Individuals prioritize mental health but encounter obstacles such as cost and scheduling with traditional therapy.

Users exhibit optimism towards an AI-powered mental health app, envisioning it as a source of convenient and accessible tools, guided exercises, self-management features, mood tracking, and personalized plans.

User Persona

On the hunch that my insights were challenging my biases and would lead me into a different direction I created a user persona to further empathize and understand the user.

The user persona created is Joseph, a 34 year old beverage director who regularly experience anxiety, negative thought spirals and restless sleep.

His attributes were directly taken from my research insights and paved the way through sitemap to the outcome of the project.

Define

Discoveries in user research influenced a shift away from the initial concept of an AI-chat feature.

Key insights influencing this were:

  • User skepticism towards chatbots

  • The point of view that a therapist app should be in addition to and not replace traditional therapy

There just wasn’t enough evidence to back up efforts put into an AI chat feature. This directed the focus instead on leveraging AI to provide users with personalized plans and evidence-based exercises to manage their mental health.

Insights that supported this direction were:

  • User optimism towards an AI-powered mental health app.

  • Envisioning it as a source of convenient and accessible tools, guided exercises, self-management features, mood tracking, and personalized plans.

Information Architecture

I started with a feature roadmap to lay down the foundation for my sitemap.

Important features included in this were:

  • Mood check-in through ‘morning preparation and evening reflection’

  • Weekly AI generated summaries based on users input through mood check-ins

  • AI generated personal plans based on weekly summary

  • Journal, CBT, breathe, and meditation exercises

  • Trends/insights showing users their input history displayed in organized charts

  • A log of all user actions

— Final Sitemap

Design

I started the branding process by creating a mood board with similar colors and design styles including graphic design, typography, UI design and general artwork. From here I established brand values and a characteristics which provided me a platform to sketch logos off of. I was able to iterate on a few sketches in Illustrator and established a logo and wordmark. From here I narrowed down on color palette and typography. As for both, the approach was to find a combination that was friendly and calming but serious enough to address mental health in a professional manner.

Color was something that was iterated on throughout the process of designing and not decided on before hand. I had to understand how it would feel while being used within the designs. This led me to continuously iterating on color though hi-fi designs and finally cementing a palette after usability tests.

Branding

Wireframes

As I started to flesh out lo-fi wireframes I came to the realization I had a very broad scope for an MVP. This led to a shift in narrowing the attention to a ‘morning preparation’ and the personalized weekly plan of a CBT exercise flow.

Early lo-fi wireframes started with all the first level navigation screens. At this point I was also still looking into ways of implementing the chat feature in a more subtle way.

— Lo-fi Wireframes

— Early Hi-fi Wireframes

An important aspect while iterating on these wireframes was to ensue visual hierarchy was prominent and that the design and usability was simple and not cluttered. The point is to maintain a calming and easy to use app rather than frustrate a user who might be suffering mentally.

— Home Screen Iterations

Test

Usability Testing

In order to test the usability with participants I first created an interactive prototype with unpolished wireframes. The goal here was to test out two main flows, the morning preparation, and the weekly CBT exercise flow in order to learn more about how users interact with them.

I conducted unmoderated, remote usability tests with 6 participants using Maze to direct them into a series of tasks.
I had the users start from the home screen and navigate to these exercises in order to evaluate their discoverability. For the morning check-in I had three different versions of the flow so the goal here was to find the most effective one.

After analyzing the results and measuring the impact/risk of user success caused by potential design changes, I moved forward with iterations.

The nature of the testing was well received though I was able to identify and execute on small improvements based on the what I learned from the usability tests. Here are some of the updates:

  • Decision to move forward with morning preparation V1 with added sleep and motivation questions

  • Implemented questions on sleep and motivation to be on two separate screens instead of one

  • I changed the language in the CBT exercise from ‘positive/negative thoughts’ to ‘useful/unhelpful thoughts’

  • Included more labeling on the exercise cards within the CBT plan to include ‘breathe’ or ‘meditation’ if these were incorporated in the exercise

  • Decreased saturation/brightness of pink accent color used throughout

  • Slightly increased saturation of primary color

Before

After

Here are some screens of the final hi-fi designs after several rounds of iteration and refinement from testing and peer feedback.

Final Designs

Next Steps

Given additional time, I aspire to refine this project further, exploring alternative approaches to information architecture and expanding activity flows. Extensive usability testing with varied objectives would be pivotal in this phase.

Summary

Challenges Faced

Overall, this project was a daunting and challenging undertaking. Navigating the saturated landscape of mental health apps and refining the information architecture to align with user needs proved to be a hefty task. Collaborating with a teammate might have offered valuable perspectives during these stages.

Looking back, I recognize that the initial solutions went through significant shifts during the ‘define’ stage. Analyzing the research and user persona led me to reconsider what’s important and prioritize features aligning with user needs. Ideally, more time would have allowed for the design and testing of a mid-fidelity prototype with the initial AI-chat flow, potentially yielding valuable insights for comparison.

What I Learned

The research I conducted on mental health and mindfulness techniques proved valuable even on a personal level. Additionally, the importance of user-centered design became very clear. Without user research, relying solely on initial ideas would likely have resulted in a product with limited relatability.

How This Project Can Inform My Future

This experience serves as a reminder of the importance of paying close attention to research and basing design decisions on its findings. It's crucial to directly associate information architecture decisions with key insights from user research and remain committed to those choices.

What I Could Have Done Differently

In retrospect, I could have improved by more closely linking information architecture choices with user research insights. During this stage, I may have strayed slightly focusing on competitors, leading to a sense of indecision.

Conclusion

Although the project initially centered on an AI-chat therapist feature, iterative design and research led to a reimagined focus and the elimination of the chat function. Despite encountering challenges, this endeavor afforded invaluable learning experiences. Given the opportunity, I am eager to continue iterating on this product and witnessing its evolution into a tangible solution.

Thank you for exploring this case study!