AI personalized nutrition: ZOE to launch app following twin investigation

15 Jun 2020 --- ZOE is set to launch a test kit and app that uses artificial intelligence (AI) to develop personalized eating plans based on a person’s unique gut microbes and dietary inflammation. The health science company has also sponsored PREDICT (Personalized Responses to Dietary Composition Trial), touted as the largest ongoing nutritional study of its kind. It has now revealed that identical twins respond differently to the same foods, showing that genetics plays only a minor role in determining personal nutritional responses. Therefore, tests offering personalized nutrition advice must be more holistic than just using genetic information, says ZOE.

“There isn’t a one-size-fits-all nutritional solution, so the food and health industries need to tailor their products if they want to optimize health via what we eat. To tackle the obesity pandemic at a policy level, it will become important to understand how these inter-individual variations in response to food translate into higher or lower levels of hunger, eating after a meal and into weight gain, maintenance or loss,” Dr. Ana Valdes, a study author, ZOE-collaborator and a professor at the University of Nottingham, UK, tells NutritionInsight. 

ZOE is touted as offering personalized eating plans to help people hit their healthiest weight by reducing dietary inflammation and supporting the gut. The app already has a 20,000 strong waitlist and Dr. Valdes notes that it will be available in the US in early August, with the UK following in fall. 

“The product appeals to those seeking to improve their health, be it losing weight or reducing their risk of long-term health issues. There’s currently nothing on the market that gives you such an in-depth scientific view of your holistic nutritional response. This is the best science you can get to understand how your body works and then have the ability to alter your response to food through our app. The ZOE app offers personalized food scores that teach you the best food for your body and how to pair foods so that eating healthy isn’t restrictive,” explains Dr. Valdes. 

 The app already has a 20,000 strong waitlist.Notably, the company is also the creator of the COVID Symptom Study app, which uses AI and symptom data to predict the spread of COVID-19 in real-time. The app currently has 4 million users globally and confirmed that people who are obese are more likely to end up in the hospital with COVID-19, as confirmed by a Public Health England data review. 

However, ZOE is not the only personalized nutrition technology available to consumers. US-based Baze recently received approval from the US Food and Drug Administration (FDA) for its at-home blood collection process, which allows customers to assess their blood nutrient status. Meanwhile, Mako Medical Laboratories has launched a genetic test targeted at nutrition performance and customized meal plans, created to help consumers know exactly what their body needs and what foods should be eaten in order to optimize performance.

Weight management via reduced dietary inflammation 
The PREDICT-1 results, as published in Nature Medicine, reveal a wide range of inflammation responses after eating, even among apparently healthy people. Dietary inflammation is linked with increased risk for conditions such as heart disease, Type 2 diabetes and obesity. The study suggests improved weight management and health could be achieved by eating food that is personalized to reduce dietary inflammation. Other key findings of PREDICT-1 include:

• Identical meals eaten on different days showed that individual responses to the same foods were remarkably consistent for each person, despite wide variation between participants. 
• There is no fixed “perfect” time to eat for nutritional health, with some people clearly metabolizing food better at breakfast while others saw no difference. 
• Optimal meal composition in terms of macronutrients is also highly individual, so prescriptive diets based on fixed macronutrient ratios are too simplistic and will not work for everyone, according to ZOE. This means that a sensitive glucose responder may need to reduce carbohydrates, whereas someone else may be able to eat these freely, for example. 
• The relationship between the calories consumed in a meal and nutritional response is weak.
• The form a food is in – cooked, chopped or ground, for example – will produce a drastically different result. 

The researchers’ main goal with this research was to understand scientifically, in large-scale, high-quality studies, the links between what we eat, our immediate responses to food and our health. “We didn’t expect genetics to be such a weak predictor, nor did we expect meal timing and sequence, along with individualized responses to be major contributors,” Dr. Valdes details. 

“We are only now beginning to understand quantitatively how to modulate our health via nutrition. There are likely to be limits to how much we can impact our health, but right now our greatest limitations are measuring things in real-time such as insulin and triglyceride responses at home, in the same way that we measure glucose,” she adds. 

Measuring a wide range of markersGenetics have a negligible effect on triglyceride levels.
The study was led by Professor Tim Spector and his team at King’s College London and ZOE, in collaboration with Massachusetts General Hospital and additional researchers in the US, UK, Italy, Spain and Sweden. It recruited 1,103 participants from across the UK and US, including 660 identical and non-identical twins from the TwinsUK cohort. The study measured a wide range of markers from blood glucose, fat and insulin levels to exercise, sleep and gut bacteria. 

Each participant was able to eat scheduled set meals and their own free choice of foods while wearing a continuous glucose monitor and activity tracker throughout the duration of the study. They also took finger-prick blood samples to monitor blood fat levels and collected stool samples for microbiome analysis.

Forthcoming scientific papers stemming from PREDICT-1 include findings on the factors that contribute to hunger and energy lulls, and post-meal fat response and its influence. 

Last year, the research team also found that individual responses to the same foods are unique, even between identical twins, demonstrating that one-size-fits-all dietary guidelines are too simplistic.

PREDICT-2 studied 1,100 UK and US participants, concluding in March with the initial findings set to be published later this year. The study included complex microbiome profiling, which has led to the discovery of a specific microbe that may determine people’s ability to metabolize food better. Additionally, the next stage of the PREDICT studies will be announced soon. 

“Modulating appetite and energy expenditure through how we eat, as well as the sequence and time of the day at which we eat, are two of the ambitious goals that we have set out for the near future. We are also very interested in understanding the ways in which what and how we eat can improve our immune response to viral or bacterial infections. Therefore, we are focusing on links between food, the gut microbiome, and inflammation and the innate immune system,” concludes Dr. Valdes. 

By Katherine Durrell