All Projects

Receipt2Nutrition

The world’s first large-scale voluntary panel that contributes digital receipt data from loyalty cards anonymously to assess the data’s potential to predict dietary behaviour automatically

Motivation

  • Problem: Diet is a major modifiable risk factor for chronic diseases. Chronic diseases cause more than 70% of deaths worldwide and 85% of deaths in Europe, resulting in 80% of healthcare spending in Switzerland (WHO 2016, FDHA 2018). In addition, roughly one-third of the food produced for human consumption is lost or wasted globally. This overproduction of approximately 1 billion metric tons of food, which is never consumed, uses up 30% of the world’s agricultural land, 20% of freshwater consumption and is responsible for roughly 8% of the anthropogenic greenhouse gas emission (FAO et al. 2018).
  • Solution approach: Automated and scalable monitoring and interventions can counteract these alarmingly increasing figures, be it in the contexts of healthy diets or sustainability. Digital receipts automatically captured through loyalty cards enclose relevant information that can help understand consumer beahvior and offer personalized self-monitoring on the individual-level. Similar to the investigations on dietary behavior, the framework is used to predict food waste behavior – in order to identify the quantities that are wasted and that are actually consumed.
  • Framework: Switzerland offers a quite unique milieu: the loyalty programs offered by Coop and Migros, the two main grocery stores, reach more than 3 million of households each, and 80% of the retail revenues are associated to loyalty cards.
  • Purpose: The aim of this project is to develop a machine-learning model that reliably predicts an individual user’s dietary pattern from their automatically captured receipts.
  • Application: Similar to “Cumulus Green” of Migros, better understanding of users’ diets and personalized feedbacks are intended to increase awareness and to facilitate nutrition interventions and monitoring. In addition, the research on food waste behaviour could lead to novel approaches on food waste monitoring, or interventions accordingly.

Description

The Receipt2Nutrition research project aims to develop a novel and scalable approach for delivering individual-level dietary monitoring by applying data science on automatically captured digital receipts.

  • Digital Receipts: Shopping data used only includes product names, prices, and date of purchase. Hence, sales data first needs to have nutritional facts. For this, Auto-ID Lab has developed and maintained a product ingredient database, which now contains more than 30’000 most frequent Swiss products.
  • FFQ:  Sales data enriched with nutritional information is compared to the actual dietary intake. The latter is obtained through a Food Frequency Questionnaire (FFQ), which investigates into participants’ diet in the previous month for elaborating the usual dietary intake.
  • Personas: anonymous users’ descriptions (socio-demographics, shopping and dietary behavior) help understanding more patterns for developing a reliable model.
  • Data Collection: The data is collected through the BitsaboutMe platform, where participants sign up, link their data sources (Coop, Migros), and register to the study.
  • Anonymous identification: By sharing Migros and Coop shopping data, only product names, prices, and their dates of purchase will be forwarded anonymously. That information will be linked through an anonymous identifier to a diet survey (FFQ).
  • No data is shared externally: All data will be stored at ETH Zurich for the purpose of research only. Only affiliated researchers can conduct analyses for the purpose of this study. All re-personafiable data attributes are removed, i.e. timestamps, locations. No names or emails are stored. No data is shared with external third parties.
  • Participation incentives: Upon completion of the diet survey, participants receive a report assessing their current diet and a personalized Health-Shopping-Widget is activated on their BitsaboutMe dashboard. This feature will track the degree of healthiness of their future grocery baskets.

FAQ

  • Who is involved in this project?
    • This research project is conducted by the Auto-ID Lab at ETH Zurich (Chair of Information Management, Prof. Elgar Fleisch), together with researchers from the University of Zurich and the Unvirsity of St.Gallen.
    • This research project is led by the Auto-ID Lab at ETH Zurich (Chair of Information Management, Prof. Elgar Fleisch).
    • BitsaboutMe, a Swiss Start-Up located in Bern and operating in the Personal Data Economy, provide the platform to collect the data.
    • Other partners helped the realization of  the product ingredient database: GS1, CodeCheck, FoodRepo, OpenFood
  • What is the purpose of this study?
    • Receipt2Nutrition is a research project of ETH Zurich together with the University of Zurich and the University of St.Gallen in which loyalty card shopping data is analyzed in order to develop a model that predicts and monitors diets and food waste behavior.
  • Why should I participate in this study?
    • Your participation supports the realization of this project, an important step towards preventing diet-related diseases – one of the largest challenge of the 21st century.
    • As a reward, upon completion of the diet survey, you will receive a report assessing your current diet and your personalized Health-Shopping-Widget will be activated on your BitsaboutMe dashboard. This feature will track the healthiness of your future grocery baskets.
  • How can I participate in this study?
    • Follow the simple steps illustrated in this tutorial www.autoidlabs.ch/receipt2nutrition-how-to-join – it will only take ca. 30 minutes and you will also receive a personal diet assessment and a new tool for tracking the healthiness of your next grocery shopping baskets.
  • What will I get by participating in the study?
    • As a reward, upon completion of the diet survey, you will receive a report assessing your current diet and your personalized Health-Shopping-Widget will be activated on your BitsaboutMe dashboard. This feature will track the healthiness of your future grocery baskets.
  • How will you handle my data?
    • Your privacy is important to us: The Receipt2Nutrition Team only processes survey and shopping data in anonymous form (e.g. without location, timestamps, personal identifiers)
  • How can I stay informed about project updates?
    • Sign up to our newsletter to receive interesting updates about this project: click here to sign up
Partners
Working Group
Contact

Klaus Fuchs
Chair of Information Management
ETH Zurich
fuchsk@ethz.ch