Brief Description of Our Approach

To provide scientific evidence-based, specific and actionable information and guidance on food choices that we make every day and how they might impact our health, we had to quantify a relative level of positive or negative effect for every individual food item in our database as they relate to each specific illness or health concern tracked in our system.

As a foundation for data in our system we take multi-year multi-center prospective studies conducted and/or funded by the USDA, HHS, NIH and other respective agencies around the world (e.g., Framingham Heart Study). These studies are done on many thousands of people and monitor morbidity and mortality over decades while periodically collecting the information on their diet, plasma levels of markers of different diseases and known nutrients.

Here is a very brief description of our approach to creating our knowledgebase:

  1. We maintain detailed nutrition information on every individual food item in our system. Most of such data is available from the U.S. Department of Agriculture Food Database. Additionally, some data (e.g., Mercury, Gluten, Purine, Myo-inositol, Tyramine) we collect from peer-reviewed sources, US EPA, NIH agencies, and major universities.
  2. When new evidence-based data becomes available on health benefits (or adverse impact) of a specific food item or nutrient, relative to a specific illness, we have our science team review it in the context of our knowledgebase on this subject to-date, and reach out to other experts as needed. Upon confirmation, we capture and make use of such data in our algorithms which immediately gets integrated into our consumer-facing technology interface.
  3. Some foods or nutrients must be tightly controlled if one is taking certain medications (e.g., alcohol and numerous drugs; green vegetables and anticoagulants; large amounts of tyramine and Isoniazid). We maintain and incorporate such data in our algorithms. We use FDA guidance documents as the source for this food-drug interaction data.
  4. If a given nutrient is demonstrated to have a positive effect on a health condition (e.g., Vitamin A is demonstrated to reduce night blindness), then all food items that are rich in that nutrient (Vitamin A) are given a positive/higher rating as they relate to that condition (night blindness). Similarly if a given nutrient has adverse impact on a health condition (or a medication) then all food items that contain that nutrient are given a negative/lower rating relative to that condition.
  5. Our algorithms capture the amount of nutrient per typical serving and take that into account in making the recommendations (e.g., the one that is richer in Vitamin A will get a higher score than the one with less Vitamin A in the above example).
  6. Some nutrients are found to be much more significant (e.g., Vitamin A) than some other nutrient (e.g., Zinc) as they relate to a given health condition (e.g., night blindness). Our algorithms distinguish between the two (by placing more weight on the Vitamin A-rich foods as opposed to Zinc-rich foods).
  7. Certain nutrients facilitate absorption of another nutrient (e.g., Vitamin D facilitates absorption of Calcium). Our algorithms take this into account. For example, adequate Calcium intake is essential for prevention of osteoporosis. Then all food items that are rich in Vitamin D are given a more positive score as they relate to osteoporosis.
  8. Similarly some substances may reduce absorption of or increase the need for another (e.g., Caffeine interferes with calcium absorption). In the above example, all food items that contain Caffeine will receive a negative/lower rating as they relate to osteoporosis.
  9. When a relationship between a nutrient and a health condition is stronger and reproduced in multiple independent well-designed studies (i.e., the relationship between intake of foods rich in saturated fat and risk of cardiovascular disease), our algorithms take this into account by assigning a positive weighing factor and give this relationship a higher score.
  10. The process and the steps mentioned above are automated by our technology. At the conclusion of the process, there is a single score that represents the degree of positive or negative effect of any food item on each health condition (or medication) maintained in our system. These scores are the basis for all the dietary guidance produced by our algorithms.
  11. Similarly, a subset of the process described above is applied to various activities, exercises, and life style choices (e.g., drinking, smoking, acupuncture, yoga, sleeping) to establish the degree of positive or negative effect and produce a score for that activity as it relates to a health condition.
  12. For multiple conditions, the sum of these scores drives the ranking of the food items or activities and the final recommendations. Many of the conflicts in dietary and activity considerations for multiple conditions are resolved in this manner.

It is important to note that we are not making any new scientific claims, nor do we suggest that we have the ultimate approach. Our goal is to simply provide a significant improvement over the status-quo. No human being or health care specialist can properly and fully take into account the enormity, complexity and contradictions inherent in the interrelationships of food, health, genetics, medications, environment, exercise, lifestyle, etc. that affect our wellbeing. We have merely attempted to use the power of technology to produce significantly better and more relevant information to maintain healthier living. Importantly, such advice is urgently needed when many patients and consumers make food and lifestyle choices and decisions every day, in real time. The technology we have developed has the ability to provide such actionable real-time advice.