JMIR Publications

ABSTRACT

Background:

User acceptance and trust play a critical role in decision-making, particularly in domains such as personalized nutrition. This paper investigates the design and implementation of a food recommender system and explanation interfaces that provide insights into individualized food recommendations.

Objective:

The objective of this study is to examine the significance of user acceptance and trust in supporting decision-making and explore the effectiveness of explaining food recommendations in a retail-controlled service environment.

Methods:

A mixed method, user-centered design approach was employed, involving 26 participants and expert feedback from 2 professional dietitians. The proof-of-concepts were tested through deployments at 2 large companies with 45 and 16 participants, respectively. An updated demonstrator was deployed at a third large company over a 7-week study duration with 136 participants.

Results:

Despite a mismatch between participants’ food preferences and their individual healthy recommendations, the mobile application successfully explained the reasons behind the recommended meals with clear and adequate explanations. This explanation process led to an increase in trust in the recommendations. The paper discusses the design goals of the food recommender system, the challenges faced during real-life deployment in a retail-controlled service environment, and provides reflections for future food recommender systems.

Conclusions:

The study highlights the importance of explaining food recommendations in fostering user trust, even when there is a discrepancy between user preferences and healthy recommendations. The system effectively provided clear and adequate explanations, resulting in increased trust in the recommendations. The paper also discusses design goals, deployment challenges, and offers recommendations for future food recommender systems.