Prototyping

Interaction design

Design systems

GlucocareAI

GlucocareAI

Lifestyle management reimagined with the use of AI saving 2 hrs daily

Timeline

12 weeks (Jan 2024 - Apr 2024)

Team

UX Lead, UX Designer, UX Researcher

My role

UX and Project Lead

Tools

Figma, Zoom

In a nutshell

97.6 million

97.6 million

Americans had prediabetes as of 2021.

Americans had prediabetes as of 2021.

Americans had prediabetes as of 2021.

Logging every meal is tiring!
I wish it was easier

I only get boring and non-feasible suggestions for meal plans.

Logging every meal is tiring!
I wish it was easier

Ideating an AI-based solution for scanning meals easily

Ideating an AI-based solution for scanning meals easily

Ideating an AI-based solution for scanning meals easily

Followed Apple’s design guidelines

Followed Apple’s design guidelines

9.2/10

usability rating

9.2/10

usability rating

Conducted user interviews

Users can scan food easily and also get personalized recommendations

Users can scan food easily and also get personalized recommendations

Users can scan food easily and also get personalized recommendations

Impact

9.2/10

satisfaction rating

2

hours saved daily

Problem space

Despite affecting millions, prediabetes is often overlooked!

On exploring the work done in the healthcare sector using AI, we realized that diabetes management was an emerging topic. On reading articles about diabetes management and statistics on diabetes, we noticed prediabetes - a health condition leading to diabetes which is often ignored, had very little research.

352,000

young Americans diagnosed with diabetes as of 2022.

352,000

young Americans diagnosed with diabetes as of 2022.

352,000

young Americans diagnosed with diabetes as of 2022.

97.6 million

Americans had prediabetes as of 2021.

97.6 million

Americans had prediabetes as of 2021.

97.6 million

Americans had prediabetes as of 2021.

Research

For understanding the problems related to prediabetes and current research, we conducted a literature review followed by a competitive analysis. We later conducted interviews with 3 participants to understand the problem more deeply.

Literature review

Studied 15 research papers focusing on prediabetes management and role of AI in diabetes management.

We read a total of 15 research papers focusing on topics including diabetes management and AI, prediabetes management and methods of reversing prediabetes.

Five out of nine papers concluded:

Prediabetes can be reversed with lifestyle changes and healthy lifestyle management.

Five out of nine papers concluded:

Prediabetes can be reversed with lifestyle changes and healthy lifestyle management.

Competitive analysis

Studied 8 competitor apps based on fitness management or lifestyle management.

No app focusing on prediabetes and no personalization for lifestyle restrictions.
No app focusing on prediabetes and no personalization for lifestyle restrictions.

The team conducted a thorough competitive analysis of 8 applications in the domain of fitness or diabetes management. I studied 3 apps: LoseIt Calorie Counter, MyDiabetes and CaloScanAI.

Key insights

  • None of the 8 apps studied focused on prediabetes.

  • Most diabetes management apps focused on daily glucose monitoring - which is not a requirement for prediabetes

  • Regular fitness apps used for lifestyle management did not account for restrictions due to prediabetes.

  • The food scan feature was either for food or barcode and not both.

User research

Conducted interviews with 3 participants to understand their pain points.

Meal preparation or personalization aspect was missing in the current solutions.
Meal preparation or personalization aspect was missing in the current solutions.

To understand what people diagnosed with prediabetes do to manage their condition, we conducted interviews with three prediabetes patients.

Key insights

  • There was a lack of personalization in the currently existing apps.

  • Participants with no experience with using apps for prediabetes management expressed their concern with logging everything since it was a hastle and often got things wrong and said they couldn’t trust a smartphone app to manage their healthcare.

Problem definition

Brainstormed the pain points, defined objectives and designed a user flow for the solution.

After the thorough research, we brainstormed on identifying the major pain points that were not addressed.

Ideating the solution

The team performed a brainstorming session where we noted down our ideas on post-its and sorted them according to the requirements.

How might we

leverage AI to provide personalized meal and fitness recommendations, making prediabetes management easier and more efficient for users without overwhelming them?

Objectives

  • Personalized meal suggestions
    Personalizing meal and exercise plans based on the user’s habits, goals and preferences.

  • Make meal logging more efficient
    Make scanning food easier and convenient for users to track.

  • Allowing users to customize the recommendations
    Making a system that learns from user input and enhances the personalization.

User flow

We then designed a user flow that captured all aspects of the features we aimed to include.

Design

Wireframes

Designed wireframes to represent the information architecture

With the flow designed, we started designing wireframes to establish a clear understanding of the app’s flow and structure.

Visual design

Followed Apple's design guidelines and created a design system for uniformity in the designs.

We followed a design system to ensure uniformity in the application. I was responsible for creating components in the design system. Our design system was aligned with Apple’s Human Interface guidelines ensuring the color, typography and layouts were in compliance with the guidelines.

I also designed a logo for the application that would represent the app’s purpose.

High fidelity design

Designed an interactive and detailed UI with an element of AI in it.

With a solid design system in place, the project moved to high fidelity prototyping, focusing on refining the visuals and interactions. This phase involved designing with a higher level of detail and representing the use of AI in the app.

I designed the onboarding screens, the home screen, the weekly plans screens.

Weekly plans for meals and exercise

Users can plan ahead while buying groceries and planning their days

Home screen summarizing daily progress

Users can see current progress and upcoming meals at a glance

User testing

Tested the high fidelity designs with 3 participants via Zoom.

The high fidelity designs were tested with three participants - one pre-diabetic, one diabetic and one non-diabetic. The interviews were conducted via Zoom and the users performed tasks based on the apps functionality. We asked the participants questions after each task to gain feedback.

Positive feedback

Users really liked the scanning feature and the meal suggestions made planning and tracking easier!
Users really liked the scanning feature and the meal suggestions made planning and tracking easier!
  • The users liked the scanning feature and mentioned that it would be really handy and useful.

  • The concept of a smart health planner was intriguing for the users and they liked the personalization and smart recommendations.

  • The participants found the visualization feature motivating and useful.

The meal preparation ideas seem a great feature! I'm sure this app will help me and other prediabetic patients manage lifestyle more easily.

Participant 1

Diagnosed with Prediabetes 2 years ago

The meal preparation ideas seem a great feature! I'm sure this app will help me and other prediabetic patients manage lifestyle more easily.

Participant 1

Diagnosed with Prediabetes 2 years ago

Areas of improvement

Users wanted a fallback for when the AI went wrong
Users wanted a fallback for when the AI went wrong
  • A few design flaws like lack of back button were noted.

  • A participant expressed their concern when they want to scan food with multiple ingredients that are not separately visible: the system might fail.

  • Some elements in the UI had issue with visibility and readability.

Redesigning based on the feedback

Redesigned high fidelity screens to accommodate user needs.

Considering the feedback from the users, we redesigned a few aspects of the high fidelity UI by adding some elements like a more engaging AI interaction, allowing users to edit certain aspects while meal tracking, improving the UI visually, etc.

What if I have food with ingredients not visible separately? Would the AI handle that too?

Participant 3

Diagnosed with Prediabetes 6 months ago

What if I have food with ingredients not visible separately? Would the AI handle that too?

Participant 3

Diagnosed with Prediabetes 6 months ago

Allow users correct potential mistakes

Fallback for when AI might go wrong - provides users with choice and flexibility.

Additionally, we also added some functionalities like uploading medical reports for more accurate and personalized plans, enhancing feedback mechanisms and increasing AI transparency by providing information about the recommendations

Increase AI transparency

Increase AI transparency

Ensures trust and credibility in AI-suggested plans

Final deliverable

After iterating on the feedback received during testing, changes and improvements, a final prototype was designed.

Impact

3

users

3

users

9.2/10

satisfaction score

9.2/10

satisfaction score

2

hours saved daily

2

hours saved daily

  • Tested the prototype with three users and achieved an overall rating of 9.2/10 for the application’s usability.

  • On average, users saved 2 hours per day in meal planning and logging activities using the app compared to traditional methods.

The GlucoCare AI project team did an exceptional job addressing the needs of individuals with prediabetes. Their use of compelling statistics highlighted the project's urgency, and their thorough review of existing apps and academic papers added depth to their work. Overall, the team effectively conveyed the problems associated with prediabetes and articulated a promising health management solution. Their dedication and effort are truly commendable.

Dr. Min Kyung Lee

Ph.D. Carnegie Mellon University

The GlucoCare AI project team did an exceptional job addressing the needs of individuals with prediabetes. Their use of compelling statistics highlighted the project's urgency, and their thorough review of existing apps and academic papers added depth to their work. Overall, the team effectively conveyed the problems associated with prediabetes and articulated a promising health management solution. Their dedication and effort are truly commendable.

Dr. Min Kyung Lee

Ph.D. Carnegie Mellon University

Learnings and outcomes

  • Feedback-based designing
    As we tested the prototype with users, we realized how important it was to understand the user’s perspective and modify the app according to user needs.

  • A strong base research goes long way
    Since we researched and chose a niche topic, we were able to focus on the target group and their requirements and design for them.

  • Designing for a smaller group poses challenges for testing
    Since the app was designed for a specific group of users, it was difficult to test it with real users and get feedback.

To summarize

Designed a meal preparation and logging app for prediabetic patients with an average user satisfaction rating of 9.2/10.

Despite multiple applications for fitness and diabetes management and monitoring, there are no applications focusing on prediabetes. We designed an application focusing on the special group to help them manage prediabetes via lifestyle management. The primary features of the app were food logging using AI scan and providing meal plans and exercises tailored to the user’s preferences and restrictions. We received an average user satisfaction rating of 9.2/10.