BACKGROUND
01
Research
Visually impaired individuals are more prone to malnutrition due to accessibility challenges as indicated by previous research.
02
Visually impaired people have reported difficulty shopping for, eating, and preparing meals.
03
Food packaging often relies on visual information, creating challenges for visually impaired individuals. AI system that can read aloud product details and detect freshness of grocery can significantly improve their life.
BACKGROUND
Problem Statement
Visually impaired people have difficulty checking food expiration dates and monitoring food freshness.
Difficulty finding food with nutritional value.
Difficulty figuring out recipes.
SOLUTION
Prototype
SOLUTION
Demo Video
METHOD
Survey Design
Study Type
Single Factor Design
Participants
32
Independent Variable
Use of FridgeBrain App
Dependent Variables
-
Perceived helpfulness of food storage listing.
-
Perceived helpfulness of expiration alerts.
-
Perceived helpfulness of nutritional suggestions.
-
Perceived helpfulness of recipe suggestions.
-
Perceived helpfulness of food scanning.
-
Perceived helpfulness of the voice interaction assistant.
-
Perceived helpfulness of insights provided by the AI assistant.
-
Overall satisfaction with the app.
PROCEDURE
Survey Structure
5
Demographic
2
Preference
8
User Goal and Usage
Questions
Demographics / Preferences
1
Gender
2
Age
3
Frequency of Storing Food in the Fridge (past month)
4
Frequency of Finding Expired Food in the Fridge
5
Frequency of Cooking
at Home
6
Difficulty Managing
Nutritional Value
7
Trust in AI Assistance
Measurements - Helpfulness Level
1
Food Storage Listing
2
Expiration Alerts
3
Nutritional Suggestions
4
Recipe Suggestions
5
Food Scanning
6
Voice Interaction Assistant
7
Insights Provided By the
AI Assistant
8
Overall Satisfaction
Results
Demographics and Averages
Linear Regression
Conclusion
Average Helpfulness/
Satisfaction Ratings:
-
Range: 3.66 – 4.28 ✅
Linear Regressions:
-
Low correlation
-
Negative slopes: 4 (food scanning, expiration alerts, recipe suggestions, nutrition suggestions)
-
Positive slopes: 3 (food listing, AI assistance, AI voice)
-
Some individuals had high ratings for food/nutrition related issues vs. low-medium ratings for helpfulness or satisfaction of features
Limitations
-
Prototype
-
Basic structure and minimal interactions
-
No actual AI implementation
-
No screen-reading/accessibility capabilities
-
-
Testing environment
-
Not an accurate environment
-
Demo video only, no interaction or tasks
-
Next Steps
-
Interview survey participants
-
High ratings for food/nutrition related issues vs. low-medium ratings for helpfulness or satisfaction of features
-
-
Interview visually impaired and blind individuals
-
Refine prototype
-
Further develop Figma prototype
-
Incorporate AI or better mimic AI features
-
Incorporate or mimic screen-reading/accessibility features
-
-
User testing
-
Create and conduct user tests on refined prototype
-
-
Further statistical analysis using new data
