CREATING A FLAVOR-MATCHING SYSTEM WITH MACHINE-LEARNING TO INCREASE CONSUMPTION OF HEALTHIER FOOD CHOICES
CLOUD COMPUTING & BIG DATA
Author: Cynthia Kirkeby
Coauthor(s): Anthony Ko, Mohamed Abdel Latif, Matt Evilsizor, Natalie Sweis
Status: Work In Progress
Funding Acknowledgment: Workspace provided by UCI Applied Innovation’s Wayfinder Program
Reducing diabetes factors with machine-learning flavor-based customization of food choices with local, organic, and sustainable produce.
1. Increasing the consumption of ripe, nutrient-rich fruits and vegetables and the reduction of refined sugars and processed carbohydrates are at the core of a healthy eating regimen that can substantially impact the occurrence of diabetes and other obesity-related illnesses.
2. An exploration of people’s dietary behavior, with a system of big data correlated with individual feedback utilizing machine learning techniques to close the feedback loop that subsequently offers seasonal information and flavorful locally-grown food recommendations.
Despite the strong understanding within the healthcare industry of the effect, healthy eating has on the reduction of obesity-related illnesses and injuries; we have a growing population that is suffering record levels of heart failure, diabetes, and other illnesses. Part of the problem lies in the reduction of flavor levels in our produce, as it’s raised to survive transportation distances averaging more than 1500 miles.
A baseline of a participant’s eating preferences is created when they join our system. Then, we use machine learning to augment that profile with a specially curated compilation of national and state data sources to guide the participant to consume a broader range of fresh fruits and vegetables. By connecting the consumer to more local growers they will experience vine-ripened, produce that is raised for nutrients and flavor instead of transportation. The adoption of a diet with a wider range of fruits and vegetables should lead to a subsequent reduction in obesity and related illnesses within the participants.
As more participants join our system our recommendation engine will become more robust and we will be able to offer more targeted advice on new ways to broaden healthy and flavorful choices within their everyday diets. Feedback from the participants will close interest graph loops and strengthen subsequent recommendations.