ZOE is a health science company that was founded on the belief that novel digital technologies can enable human research at an unprecedented depth and scale, in order to tackle global health issues. Our approach combines artificial intelligence, digital technologies and collaboration with leading scientists around the world. Our research ranges from the ongoing PREDICT program, which is the world's largest in-depth nutritional research program, to an 800,000 person clinical trial to measure the accuracy of our AI diagnostic for COVID-19.
ZOE believes that recent technological advances make it possible to carry out high-quality human research in a real world setting, and therefore answer questions that have not previously been feasible. Combining large-scale biological data with machine learning technology and microbiome sequencing has enabled us to predict personal nutritional responses to foods and provide people with a better understanding of their unique metabolism and gut microbiome1. These personalized insights can allow individuals to make impactful changes to reduce dietary inflammation, improve gut health and improve long-term weight and health.
An overview of the problem
Non-communicable diseases (NCDs) impacted by dietary risk factors, such as type two diabetes and cardiovascular disease, are the leading cause of mortality and morbidity in the developed world, and continue to increase at alarming rates2,3,4,5. Unfortunately, current dietary and lifestyle approaches to tackle obesity and other major risk factors for these conditions, are not achieving the positive impact that we all wish for. This is also true in developing countries, where a double burden of infectious and chronic diseases exists6.
Reducing diet-related risk factors associated with these conditions is one of the most important ways to reduce the chances of ill-health, as well as lessening their detrimental impact on individuals and society as a whole7,8,9. However, our recent studies have demonstrated that our responses to food are unique, and therefore a “one-size-fits-all” approach to diet is unlikely to be the best strategy to manage weight and enhance health1. ZOE’s ambition is to improve human health through precision nutrition by understanding how to eat the right way for our bodies and harness the potential power of the trillions of microbes in our gut.
ZOE’s scientific program
Our ongoing PREDICT program of scientific research comprises the largest in-depth nutrition studies in the world to date. The PREDICT studies are designed so their data can be seamlessly combined in our machine learning models, enabling us to understand and predict personalized metabolic responses to provide people with nutritional recommendations tailored to their unique body.
By collecting biological data in unprecedented detail, from thousands of study participants, in the US and UK, on hundreds of thousands of meals, we have been able to characterize and unravel what determines the huge variability in human responses to food.
The first results from PREDICT have been published in the leading medical journal Nature Medicine in June 2020. These results have been used to create personalized “ZOE Scores” for individuals who use the ZOE product. More details about the PREDICT program can be found here.
Rethinking the future of nutrition
For many years, medication has been used as the first line of therapy for treating NCDs, yet dietary approaches have the potential to be more effective7,8,9. Nutritional science is now advancing at a rapid pace, however due to the complexity of our bodies (with our thousands of biochemical pathways) and of foods (over 26,000 chemicals have been identified in our food10), we are only starting to scratch the surface in understanding how food impacts our health at an individual level. In addition, recent research has highlighted the multiple interacting pathways linking diet, microbiome and health11,12,13,14. This has revealed how our microbiome can modulate the impact of food to transform our long term health and weight.
A big focus of ZOE’s PREDICT research program has been to explore and untangle the many interrelated and multidirectional determinants of our responses to food. This has allowed us, for the first time, to combine highly detailed analysis of thousands of gut metagenomes (all the genes of the full set of microbes in the gut microbiome), with a detailed understanding of the individual’s characteristics (e.g. genetics, circulating blood metabolites), their metabolic responses to foods and health measures1.
The metabolic response to food involves a complex chain of effects which we capture with the term ‘dietary inflammation’. Eating typical meals (which include protein, carbohydrates and fat) elicits short-term changes in blood fat and glucose levels (postprandial lipemia/ glycemia), as well as other circulating metabolites. Excessive lipemia and glycemia can overwhelm the body's normal, healthy regulatory responses, triggering a wide variety of unfavourable responses in blood lipids, rebound hypo-glycemia, immune measures and hunger. Repeated often enough, these can lead to long-term inflammation, weight gain and chronic diseases such as diabetes and heart disease15,16,17,18.
Weight gain has typically been attributed to a failure to maintain energy balance (i.e. calories in = calories out)19. However, we now know that energy expenditure is highly dynamic and individual, and that both our microbiome and the food we eat can impact our metabolism. This explains, in part, why calorie reduction diets tend to fail for most individuals and is where personalized food guidance focused on quality for each individual rather than calories could have a big impact20.
Key learnings from our research
The key messages from ZOE’s research so far can be summarized in six areas:
- Individual responses to the same foods vary
We have found dramatic inter-individual variability between how healthy individuals respond to the same foods1. For example, after a standardized high fat/high carbohydrate meal the variability between individuals in postprandial lipemic and glycemic responses was 103% and 68% respectively (coefficient of variation (CV)), amongst 1,102 healthy individuals in the PREDICT 1 study. This was substantially higher than fasting TG and glucose variability (50% and 10% CV respectively), showing that measuring an individuals postprandial response allows better discrimination between individuals in their metabolic health. Importantly these findings confirm that there is no "one-size-fits-all" approach to nutrition, and point to the need for personalized guidance in eating that takes into account an individual’s unique biology.
- Our gut microbiome plays an important role in our metabolic health21,22
We have found strong links between the microbes in our gut and our unique individual metabolic responses to food. We have also established links between specific foods and dietary patterns, and individuals’ microbes in their gut at an unprecedented resolution21. Further, our research shows that a diverse diet rich in unprocessed, high fibre, plant-based foods supports the growth of healthy gut microbes, while a diet low in diversity and high in processed foods is associated with microbes linked to poor metabolic responses and long-term ill health. An exciting and novel finding is that even identical twins share only 37% of the same gut microbes (compared with 35% in unrelated individuals) (unpublished data), establishing that genetics only plays a minor role in shaping our microbiome. Thus the gut microbiome is an important and modifiable target for personalized nutrition.
- ‘Dietary inflammation’ has a large impact on our health23
Dietary inflammation is the term we use to capture the complex chain of unhealthy effects that can be caused by what we eat. Our research shows that the huge variability in lipemic and glycemic responses is also associated with large variability between individuals in other dietary inflammatory measures, including immune cell parameters and atherogenic lipoproteins. Dietary inflammation is also associated with the types of microbes in our gut; with good gut microbes (a measure of good gut health) associated with a favourable response to foods (i.e. lower dietary inflammation after meals) and lower levels of body fat, especially visceral fat22.
- Weight gain is not just a result of too many calories
Our research confirms that we should look beyond calories when striving to maintain body weight. We found that different foods with the same number of calories can result in different levels of hunger and calorie consumption at the next meal which is sustained over the following 24 hour period24. We also observed differences in postprandial responses independent of the amount of calories in the meal with subsequent effects on dietary inflammation, including responses to subsequent meals.
- We are not prisoners of our genes1
By looking at the many inter-related determinants of responses to food, we have found, for the first time, that genes only play a minor role in determining our responses to food (9.6% for glucose and 0.8% for triglycerides; SNP-based), and that even identical twins respond very differently to the same foods. These findings suggest that diet and lifestyle strategies aimed at reducing dietary inflammation can be useful to improve long-term health and weight management.
- Machine learning can be used to generate personalized nutrition advice
By measuring thousands of people’s metabolic responses to standardized test meals, as well as free living meals, we have been able to use machine learning models to generate meaningful, personalized “ZOE scores” for any food or meal. These scores provide insights into the kinds of foods that may reduce dietary inflammation, improve an individual’s gut microbiome, and help them reach a healthy weight, which may improve long term health outcomes. As our studies continue, we expect these scores to further improve.
Our research has shown that there is huge inter-individual variability in our responses to food which is mainly determined by non-genetic factors including the gut microbiome. Our ambition at ZOE is to continuously take the latest discoveries from our research and make these available to the public. We will do this by providing tests and personalized ZOE scores that reflect our latest research findings, using AI models to make precision nutrition a reality.