Related Experiment Video
Updated: May 2, 2026
07:14
Virtual Agent for Real-Time Motivational Interviewing by Integrating Adaptive Nonverbal Behavior and Language Models
Published on: December 23, 2025
1.1K
A Fine-Tuned Multimodal AI Chatbot for Dietary Health and Nutrition, Purrfessor: Development and Mixed Methods
Linqi Lu1, Yifan Deng2, Chuan Tian2
1Department of Communication, University of North Dakota, 221 Centennial Drive, Stop 7169, Grand Forks, ND, 58202, United States, 1 7017772137.
JMIR AI
|April 30, 2026
Summary
Purrfessor, an AI chatbot, enhances dietary guidance by accurately recognizing ingredients and generating high-quality recipes. This multimodal AI shows promise for personalized health conversations and informed decision-making.
Area of Science:
- Artificial Intelligence in Health
- Human-Computer Interaction
- Nutritional Science
Background:
- Multimodal AI integrating food data offers potential for meal analysis and dietary guidance.
- The practical usefulness of AI for everyday dietary decisions is not well understood.
Purpose of the Study:
- Introduce Purrfessor, an AI chatbot for personalized, multimodal dietary guidance.
- Evaluate Purrfessor's performance in ingredient recognition and recipe generation.
Main Methods:
- Trained Purrfessor on diverse datasets including USDA FoodData Central and image-recipe pairs.
- Employed a session-based interaction model with a two-phase evaluation (AI and human scoring).
Main Results:
- Purrfessor achieved 0.90 cosine similarity in ingredient recognition.
- Outperformed baseline models in recipe completeness, consistency, and clarity.
- Human evaluation showed high accuracy, relevance, and clarity in Q&A, despite minor hallucinations.
Conclusions:
- Anthropomorphic chatbot design and multimodal AI can foster engaging dietary health conversations.
- AI-driven dietary guidance can support informed health decisions and personalized communication.
- This work informs the design of user-centered AI health assistants for digital health interventions.
