Gazelle

Gazelle

Elevator Pitch

Elevator Pitch

For families with young children who have food allergies, every meal can be a source of anxiety. Gazelle uses AI to take the guesswork out of allergen safety—so parents can breathe easier, knowing their child is protected.


While other tools like Fig, Amulet, and Nima focus on detection, Gazelle goes further: combining deep learning and real-time monitoring to help families avoid allergens before exposure ever happens.


Gazelle leverages context-aware assistive AI, combining natural language processing, decision logic, and behavioral learning to help families interpret food safety risks and respond in real time.

About

About

Project Type:

UXUI, Branding


Application:

Mobile App, Wearable App, Sensor


Timeline:

12 weeks

Why I need to address allergy issue?

Food allergies are significant health concerns across the world.

Food allergies are significant health concerns across the world.

Worldwide food allergy population is estimated up to

220-250 million

220-250 million

In the U.S, Food Allergy Population is estimated up to

32 million

32 million

Why should us care?

Why should us care?

Severe allergic reactions to food, known as anaphylaxis, cause an estimated 200,000 deaths per year globally.

Severe allergic reactions to food, known as anaphylaxis, cause an estimated 200,000 deaths per year globally.

Reference:

Reference:

World Allergy Organization

The Food Allergy Research & Education (FARE) organization

World Allergy Organization

The Food Allergy Research & Education (FARE) organization

Through the interviews, I discovered an inspiring story from Jenny, a mother share about terrifying incident of her 4-year old son Jason when he was eating at his friends’ birthday party.


He took in a small piece of peanut which cause him trouble breathing and hive that last for a week.


How can I prevent parents and kids from accidents like this?

Product Goal

Product Goal

Create an AI-based monitoring system that empowers families to effortlessly protect their children from allergens.

Target User

Target User

Families with children who have food allergies.


Primary users: Children with food allergies at the age under 8
Secondary users: Parents

Families with children who have food allergies.


Primary users: Children with food allergies at the age under 8
Secondary users: Parents

Persona

Persona

Ideation

Ideation

HMW

HMW

How might we build an efficient communication?

How might we build an efficient communication?

How might we make users feel secure about their allergy status?

How might we make users feel secure about their allergy status?

Features

Features

App User: parents

App User: parents

Tracking. Check kids’ allergy, health condition and GPS location. Send Request to the wearable.

Tracking. Check kids’ allergy, health condition and GPS location. Send Request to the wearable.

Chat. View pictures, text&voice message, voice&video call.

Chat. View pictures, text&voice message, voice&video call.

Wearable User: kids

Wearable User: kids

Food sensor provides food sample data.

Food sensor provides food sample data.

Provide instant feedback. Provide haptic feedback so the kids doesn’t need to wait for long when they accept food from others.

Provide instant feedback. Provide haptic feedback so the kids doesn’t need to wait for long when they accept food from others.

Hi-Fi Wireframe

Hi-Fi Wireframe

Onboarding

Onboarding

Home

Home

App Request food sample data from wearable

App Request food sample data from wearable

Wearable Send An Image

Wearable Send An Image

User-testing

User-testing

App task: Analysis

App task: Analysis

Before

After

  1. It's more practical to differentiate between sender and receiver

  1. Given that the recipient is a child, it would be more practical to set 'voice message' or 'video call' as the default options

  1. The progress indicator should be made more prominent

User-testing

User-testing

Wearable Task: Children execute food analysis requests initiated by their parents.

Wearable Task: Children execute food analysis requests initiated by their parents.

Homepage

receive request-reminder

camera

Analysis

Done

Before

After

Homepage

receive a request

Pull out sensor stick

Food SAMple instruction

Analysis

Done

  1. Capturing images remains challenging for young children and often requires additional guidance. Moreover, the accuracy of image analysis raises reliability concerns in high-risk scenarios


  2. From the wearable device's perspective, it should not permit the declining of requests from parents


  3. The child user must have someone read the text to them


What are the challenges&reflection?

What are the challenges&reflection?

Designing for kids demands greater empathy to fulfill their unique needs. Since they might not comprehend text as adults do, we simplify our visual and auditory cues for easy understanding.


Given enough time, going through multiple iterations is really worthwhile. Every time we do user-testing, it's amazing how we always end up discovering something new and learning from it.


One key takeaway was the importance of analysis transparency. Users—especially parents—need more than just a yes/no answer. I proposed a report feature to provide parents with a transparent view of how allergen detection results are generated. Although not yet implemented, this would help users better understand and trust the system’s decision-making process.


While the core detection flow was functional, I realized that edge cases and failure scenarios—such as inconclusive results, sensor malfunctions, or delayed alerts—require just as much design attention. In sensitive workflows like allergen detection, clarity in the face of uncertainty is what truly builds user confidence.


Moving forward, I would focus on expanding Gazelle’s fallback system, exploring ways to:

  • Handle inconclusive or failed detections with adaptive guidance

  • Alert caregivers immediately when irregularities occur

  • Offer safe alternatives or next steps when certainty isn't possible

Designing for kids demands greater empathy to fulfill their unique needs. Since they might not comprehend text as adults do, we simplify our visual and auditory cues for easy understanding.


Given enough time, going through multiple iterations is really worthwhile. Every time we do user-testing, it's amazing how we always end up discovering something new and learning from it.


One key takeaway was the importance of analysis transparency. Users—especially parents—need more than just a yes/no answer. I proposed a report feature to provide parents with a transparent view of how allergen detection results are generated. Although not yet implemented, this would help users better understand and trust the system’s decision-making process.


While the core detection flow was functional, I realized that edge cases and failure scenarios—such as inconclusive results, sensor malfunctions, or delayed alerts—require just as much design attention. In sensitive workflows like allergen detection, clarity in the face of uncertainty is what truly builds user confidence.


Moving forward, I would focus on expanding Gazelle’s fallback system, exploring ways to:

  • Handle inconclusive or failed detections with adaptive guidance

  • Alert caregivers immediately when irregularities occur

  • Offer safe alternatives or next steps when certainty isn't possible

User-testing

User-testing

Wearable Task: Children execute food analysis requests initiated by their parents.

Homepage

receive request-reminder

camera

Analysis

Done

Before

After

Homepage

receive a request

Pull out sensor stick

Food SAMple instruction

Analysis

Done

  1. Capturing images remains challenging for young children and often requires additional guidance. Moreover, the accuracy of image analysis raises reliability concerns in high-risk scenarios


  2. From the wearable device's perspective, it should not permit the declining of requests from parents

Homepage

receive request-reminder

camera

Analysis

Done

Dan He • 2023