PREDICT: The world's largest in-depth nutritional research program in the world

October 17, 2020

PREDICT: The world's largest in-depth nutritional research program in the world

ZOE’s approach to nutrition is underpinned by science, and a big part of that is our PREDICT research programme - the largest in-depth nutritional research program in the world.

The PREDICT studies are led by our scientists at Massachusetts General Hospital, King’s College London, Stanford Medicine, and Harvard T.H. Chan School of Public Health.

In this post, we take a closer look at:

• The science behind PREDICT

• How the results are helping us build algorithms that can predict your individual responses to food

• How to get involved in ongoing research as a citizen scientist, from the comfort of your own home.

The story starts here

Our founders - Tim Spector, Jonathan Wolf and George Hadjogeorgiou - have long believed that food and health are deeply personal, and that science is the key to unlocking the connection between the two.

The journey to PREDICT started 25 years ago, when Tim set up TwinsUK - a groundbreaking study of thousands of pairs of identical and non-identical twins, aiming to answer important questions about health, genetics and lifestyle. 

ZOE was founded in 2018 with the aim of combining the latest nutritional science with cutting edge machine learning technology to help everyone understand their personal responses to food and eat in the way that suits your body best. But humans are complicated - and so is nutrition - so we needed lots of data from lots of people, which is where PREDICT comes in.

What is PREDICT?

The “Personalized Responses to Dietary Composition Trial-1 ” (PREDICT-1 for short), was the first in a series of large-scale, robust nutritional science studies designed to quantify and predict individual metabolic responses to different foods.

More than 1000 people volunteered to take part in PREDICT 1, including 660 identical and non-identical twins from the TwinsUK cohort, providing detailed measurements covering a wide range of markers from blood glucose, fat and insulin levels to exercise, sleep and gut bacteria (microbiome). 

Each participant spent a full day in our clinic at the start of the study where they gave blood, urine and poop samples. We also made detailed measurements of their responses to standardized muffin-based meals with carefully controlled calorie, fat, protein, carbohydrate and fiber content. 

They then carried out the rest of the study at home, eating a schedule of muffin meals and their own free choice of foods (all of which had to be carefully weighed and photographed) over two weeks. 

Throughout the duration of the study they wore a stick-on continuous glucose monitor that tracked their blood sugar 24/7, showing us how their body responded to each meal throughout the day. They also took regular fingerprick blood samples to measure blood fat and other markers, and wore activity trackers to keep tabs on their sleep and exercise.

It all added up to millions of data points - and 32,000 muffins - which our researchers fed into their hungry computers, building a machine learning algorithm that can predict individual responses to any food. 

The results were published in the scientific journal Nature Medicine, and the algorithm now forms the basis of our ZOE food scores, which tell you whether a particular food suits your unique biology.

Taking PREDICT home: PREDICT 2

Our next study, PREDICT 2, was designed to be similar to the original PREDICT 1study. But instead of asking people to come for a whole day at the clinic, we wanted them to be able to do the study entirely from home, with the exception of a quick trip to a local Quest centre for a blood test on the last day.

Each participant received a pack containing everything they need for the 10-day study including:

Nearly 1000 people in the US took part in the PREDICT 2 study. They have helped us validate the at-home analysis methods that now form the backbone of our ZOE test kits and, of course, have provided more data to help improve our prediction algorithms. 

Be part of our ongoing research

Since PREDICT 1, we’ve expanded the program to include a range of further studies such as PREDICT-Carbs, which looked at individual responses to different types of carbohydrate. There’s also PREDICT-Cardio, which looked at the link between individual responses to food and heart disease risk. 

The ongoing PREDICT studies are now the biggest nutrition study globally, involving thousands of people in the US and UK.

We’ve delivered more than 60,000 muffins and collected over 4 million glucose readings, 56,000 blood fat readings and 12 Terabytes of data from microbiome analysis (that’s more data than the Hubble telescope generates each year). 

But we aren’t done yet. Our research continues beyond PREDICT 2. Why? Well, the more data we collect, the more we can learn about our individual responses to food and how they impact our health. 

You can choose to take part in the next phase of our research by signing up for our at-home ZOE test kit - which is based on the PREDICT 2 study pack.If you meet the eligibility criteria, you will be asked if you want to take part of the next phase of our research. 

Even if you can’t join the study, you’ll still get all the benefits of being able to understand your unique nutritional responses and get personalized recommendations for foods that fit your biology best.

PREDICT in bullet points

  • Well-designed, robust research studies are the only way to get definitive answers about nutrition.
  • Our PREDICT studies are designed to help us understand how and why people respond to differently to the same foods.
  • Studying how different people respond to food requires lots of people and a lot of data about their food responses and overall health.
  • Our PREDICT studies have helped us design and validate our at-home ZOE test kits. 
  • The data from our trials enable us predict how you will respond to different foods.
  • The more data we gather, the better our prediction algorithms get for everyone. 
  • Learn more about the science that powers PREDICT here.

Learn more: