An Online Multimodal Food Data Exploration Platform for Specific Population Health: Development Study

Area of Science:

• Nutrition Science
• Health Informatics
• Data Science

Background:

• Nutrient needs change throughout life, impacting overall health and disease prevention.
• Accessible, credible online food knowledge empowers healthier dietary choices for diverse populations.
• Current online resources may lack comprehensive, evidence-based dietary data.

Purpose of the Study:

• To develop an online multimodal food data exploration platform for accessing evidence-based diet and nutrition information.
• To create a user-friendly platform with a high-quality food knowledge graph for specific populations, such as pregnant women.
• To evaluate the platform's usability and effectiveness from both system and user perspectives.

Main Methods:

• Developed the "Food Atlas" online platform in collaboration with experts from the National Institute of Nutrition and Health and Peking Union Medical College Hospital.
• Constructed a multimodal food knowledge graph with 2011 entities and 23,497 images, featuring interactive visualizations and natural language/image-text retrieval.
• Evaluated the platform's usability with 28 participants across four user groups (pregnant, clinicians, dietitians) using the System Usability Scale.

Main Results:

• The Food Atlas knowledge graph contains 2011 entities and 26 relation types, offering comprehensive data and functions superior to existing platforms.
• The platform achieved a mean System Usability Scale score of 82.5, indicating above-average usability tailored to target users.
• 96% of participants found the platform highly consistent, highlighting the value of involving health professionals in developing online resources.

Conclusions:

• The developed online multimodal food data exploration platform meets the demand for accessible, credible, and evidence-based dietary resources.
• Broader implementation of such platforms can disseminate nutrition knowledge, aiding healthier food choices across all life stages.
• Further research and development can enhance the popularization of evidence-based nutrition information for public health improvement.