Project ‘Gredient’ Identifies Food Allergy Risks
Gredient is the winner of the 5th Year MIDS Capstone Award, sponsored by Databricks, for the Class of 2020.
32 million Americans have a food allergy and millions of others have dietary restrictions. To eliminate the cumbersome chore of inspecting ingredient lists for allergens, 5th Year MIDS Students Isa Chau, Silvia Miramontes, Emma Russon, JJ Sahabu, and Chelsea Shu created Gredient, an iOS app that does the reading for you.
Tell us about your project.
Isa: Gredient is an award-winning iOS app that helps users check whether a product contains ingredients the user wants to avoid. The backend is powered by an optical character recognition (OCR) model hosted in an AWS serverless cloud infrastructure. With an OCR model, Gredient can handle even the rarest ingredients that users need to avoid, instead of relying on previously compiled ingredient or product databases.
What inspired Gredient?
JJ: As consumers are becoming more health-conscious and allergy rates are increasing, shoppers can spend hours at a grocery store checking for harmful ingredients on product labels. Health-oriented, our team was inspired to address this issue and create a product that helps individuals maintain a safe and healthy lifestyle, with an app that would reduce the pains of watching what ingredients are in processed foods.
What was the timeline or process like from concept to final project?
Silvia: We completed the product within a 14-week time frame. From the start, we noted the ambitiousness of our project and understood that time would be our largest constraint. Our greatest challenges in this time frame were deciding and achieving a workable cloud infrastructure, as well as learning how to properly construct a mobile application for iOS devices.
Since we had a good sense of our technical weaknesses and strengths, during our very first meeting, we devised a strategy to maintain momentum by setting deadlines to meet every week and stay on track. We also developed a set of guidelines to follow within the team to maintain accountability. The guidelines also indicated what to do whenever any team member encountered an issue. We believe that the final product would not have been possible without the team’s perseverance to work through the hurdles of app-development.
Emma: With five members, we split our group into sub-teams to tackle front-end development and back-end modeling and infrastructure. These sub-teams spent the majority of the time constructing their respective components of Gredient, and worked to combine the front and back ends in the last couple of weeks of capstone.
How did you work as a team? How did you manage to work on your project as members of an online degree program?
Silvia: During the early stages of the course, we spent approximately 2 weeks investigating optimal ways to create our cloud infrastructure. After researching and seeking advice through our network, we decided to utilize Amazon Web Services (AWS).
Isa: Once we decided on the general architecture of our project, we split our team to specialize on the front-end and back-end, with one person as a go-between to make sure the two halves of the project were compatible and also to help wherever extra hands were needed. We also had two weekly video meetings, a goal-setting/task-delegation meeting, and a second check-in meeting halfway through the week. Our weekly goals were a big part of keeping our momentum up throughout the semester. We would also have informal “work-meetings” that any member could start or join to work together.
How did your I School curriculum help prepare you for this project?
Emma: We found W266: Natural Language Processing with Deep Learning and W207: Applied Machine Learning to be most helpful in preparing us for the development of the OCR and NLP models used in Gredient’s backend. With a serverless architecture, W205: Fundamentals of Data Engineering also proved to be helpful with the construction of data pipelines. Additionally, skills gained in W201 and W209 were helpful in designing engaging and effective presentations.
Do you have any future plans for the project?
Isa: We would love to release Gredient on the App Store, but for liability reasons, we need to seek legal counsel before doing so. In the future, we hope to develop a more sophisticated language mode to improve Gredient’s accuracy, improve the user interface; and also to develop Gredient for Android phones.
JJ: We also are interested in adding premium features such as a scan history page, the creation of multiple profiles, and a harmful ingredients list. We will eventually seek partnerships with food and grocery delivery services to give users peace of mind when ordering dinner or buying groceries that their allergens are protected.
How could this project make an impact, or, who will it serve?
Chelsea: There are 32 million people who have a food allergy in America, of which 200,000 require emergency medical care. Furthermore, grocery shopping while taking allergies into consideration can be time-consuming, nerve-wracking, and cumbersome. Long ingredient labels with fine print can make it difficult for the visually-impaired to read, and it can also cause uncertainty among people with allergies or caretakers of whether they truly read the label correctly. Our app, Gredient, aims to change the shopping experience by helping users to check ingredient labels for allergens quickly and accurately, so they can shop confidently and safely, and ultimately live a healthier, safer, and easier life.
