Nutrino developed a nutrition insights platform that leverages techniques from optimization theory, machine learning and big data to better understand the food that we eat and how it affects each and every one of us on an individual basis. Our platform does data analytics on three types of data: data about foods, data about individuals and the scientific literature.
We typically think of food as a simple object. But in reality, food is quite complex. The standard nutrition label in the U.S. includes about ten nutrients*. The nutrition database of the U.S. Department of Agriculture, which is considered to be the golden standard nutrition database for elementary foods, includes about 150 nutrients. However, studies have shown that there are more than 10,000 nutrients and chemicals in foods that are either beneficial or harmful for our health.
In Nutrino we divide food into four categories: elementary foods (like an apple), packaged foods, recipes and restaurant dishes. They are intentionally mentioned in this order, as this is usually their order of complexity. Our patent-pending technology can truly understand food, with all of its complexity. Our knowledge of foods constantly grows both horizontally (by knowing more about the composition of each food to nutrients and ingredients) and vertically, by scaling up the number of foods are system is familiar with. To date, our system analyzed foods from over 100,000 sources, where a source can be a food manufacturer (for packaged foods), a blog (for recipes) or a restaurant menu (for dishes when eaten out).
Our engines crawl the Internet to learn about new foods constantly. They use NLP (Natural Language Processing) and mathematical models to add additional layers of metadata on every food item. Our algorithms uses partial data to estimate nutritional values, even when the data was not previously available (a typical situation with restaurant menus). We take precision seriously, and every nutrition estimation includes error bounds. Our engine adds dozens of labels on each food item, to know for which individuals the food is suitable – whether they have food allergies, special dietary needs or simply wish to eat healthier.
The Nutrino platform continuously aggregates and analyzes anonymous information about individuals. To date, our system studied over one million profiles, including their eating habits, food preferences, dietary needs, health goals, lifestyle and data from wearable devices. Our system uses machine learning to better understand what people like to eat, and more importantly, what each individual should eat to better achieve his or her health goals.
After discovering insights on individuals, foods and scientific papers, the Nutrino platform can suggest foods for each individual. Our food suggestions are based on three pillars, that nutrition should be science-based, personalized and contextual.
Science-Based: We cannot overemphasize the need for a science-based approach in nutrition. Today, there is an estimate number of 700 diets available in the market. More than 10,000 scientific papers about nutrition are published in the English language alone every year, and tens of thousands of nutrition books are flooding the market, making healthy nutrition more confusing than ever before. Unfortunately, many of the diets, papers and books on nutrition contradict each other. In Nutrino, we carefully estimate how reliable each scientific paper is. Every statement or suggestion in our platform is always referenced to provide full transparency to our customers.
Personalized: Each and every one of us is unique, and we should also eat this way. Our food suggestions algorithms aspire to push the boundaries of tailored nutrition, by factoring in thousands of different factors about the food and the individual. Nutrino uses machine learning to always get better as you go along, and learns from both implicit and explicit user’s interactions. Our connectivity hub can power up dozens of wearable devices with real-time actionable nutrition insights and suggestions.
Contextual: We saved the best for last. Nutrino doesn’t only take into account your needs and taste, it also takes into account the context in which you make your food choices. Our algorithms take the data flowing in from your wearable devices, activity trackers and geolocation to find the foods that are most relevant for you based on your recent activities and location. When eating in, Nutrino will suggest healthy recipes to fit your needs, and when eating out Nutrino will analyze the menus of the restaurants around you to identify the healthiest options nearby.
* These are the calories, the 8 main macronutrients (protein, carbs, sugars, fat, saturated fat, trans fat, dietary fiber), and two micronutrients (sodium and cholesterol). There are typically also four micronutrients that are described on packages as a percentage of their recommended daily allowance for a standard 2,000 calories diet (vitamin A, vitamin C, calcium and iron).