Approved Research

Using genetic information to improve the prediction of food choices: A machine learning-based classification approach

Lay summary

Aims: To investigate whether genetic information can help explain individual food choice behavior? If so, to what extent the prediction accuracy of individual choice can be improved by such an integration?

Scientific rationale: Food choice and intake have profound impacts on consumers' dietary, nutritional, and health outcomes. Not only does a better understanding of the determinants of individuals' food choices serve as the foundation for accurate prediction of their food consumption behavior, but it may also facilitate the ongoing development of precision nutrition interventions. A large body of evidence from economics, marketing, and related fields has shown that product attributes (e.g., price and nutrition contents) and individuals' socioeconomic and demographic characteristics (e.g., age, gender, education, and income) are the most significant predictors of their choice behavior. Yet, recent breakthroughs in genetics and nutritional science have identified a number of novel genetic variants that are robustly associated with individual food intake and dietary patterns. This highlights the possible role of genetic factors as one of the critical determinants of individual food choice decisions. However, despite the potential importance of genetic determinants of individuals' food choices, the application of genetic factors in predicting individuals' decision-making behavior is still scarce. The paucity of individual-level genetic data combined with food consumption behavior data and communication barriers across disciplines render this line of research extremely challenging. This project attempts to fill the gap.

Project duration: 36 months.

Public health impact: This project is expected to quantify the importance of genetic factors in the individual decision-making process, broaden current understanding of how individuals make food choice decisions, and provide a more cost-effective framework for predicting human decisions/behaviors that can be used in developing precision health interventions.