author: Dov Kruger
This project is created in order to create a single unified database of nutrition to facilitate eating healthily. People with kidney stones need to track oxalates and make sure they are getting enough calcium. People with diabetes need to be able to track sugar and glycemic index. People on various diets want to track calories and nutrition. This project aims to collate all this data into a single clean JSON file for foods.
Along the way in my research I have also found that some foods can be dangerous in quantities that could conceivably be eaten, and so warnings should be available. For example Brazil nuts are rich in Selenium, but so rich that 4 a day could put someone over the upper limit, so 8 a day could incur serious side effects. Similarly chinese cinnamon (the ordinary, cheaper kind) can cause liver damage, and flax seeds while extremely healthy contain cyanide, and while it would take 1kg/day to be dangerous, it should generate a warning.
USDA has a great database for processed foods and nutritional content, but I could not find raw fruits and vegetables. It is obviously there, but has to be found. The Canadian equivalent has a database released as multiple csv files but it is denormalized, so code will have to be written to generate a combined query.
Harvard has a good site summarizing nutritional requirements and upper limits for many common nutrients. I also found an NIH publication that summarizes all the research including minerals not even listed on the Harvard site, with much more information on methodology than most people (including me) would ever want to know. The sources are listed in each file that uses them.
nutrition.json is an example file showing what I want to be able to store for each food item. All vitamin names are normalized to start with a lowercase v, so for example "vA". All minerals start with lowercase m for example "MCr". The nutrition file should eventually be automatically generated by reading in databases from USDA, canada, and any other sources needed in case they are missing any food items.
The problem with the USDA database is that it is denormalized, there is huge redundancy, the items are all wildly over-complicated. The problem with the Canada database is the opposite. It is so normalized that it is not readable without writing a program.
I have tried to compromise by agreeing on common, human-readable names not numbers. So in nutrition.json the idea is to refer to a vitamin as "vK" not as an arbitrary integer index. When there is a variance, instead of just having just one number, an array containing min and max can be used. So for example, if a food had a variable amount of vitamin K, this might be written:
"vK":[3,5],
The individual nutrients are defined in nutrient_metadata.json
While it is called nutrient, in fact the items in each food contain nutrients, antinutrients such as oxalates, physical quantities such as calories (kcal) and diabetic measures such as glycemic index.
This project is inspired by a number of excellent websites which are unfortunately incomplete. https://oxalate.org has some, but by no means enough foods, and it only has oxalate information, not other nutrition.
https://www.nutritionvalue.org/ has nutrition data from USDA but does not have oxalate information, glycemic index, and while it allows entering recipes, they are only stored personally, not shared.
My concept is to create a website that provides a single unified database of nutritional data, shows references so users know they can trust the data, provides a way to share recipes so a community can be built with shared solutions to common health problems, has references to source studies carefully documenting what is a reasonable claim based on evidence.
Additionally, I have been collecting advice on how to process food to maximize benefit. There is a common consensus of some basic facts, but information on specifics is scattered across many places, with many doctors and nutritionists charging a lot of money to give patients basic information.
Reputable doctors making claims (that are well-supported by many studies) that a plant-based diet drastically reduces the risks of cancer and heart disease include:
Dean Ornish Michael Greger Mark Fuhrman
There are also doctors out making claims that are extremely controversial and have been described as snake oil and who make large amounts of money based on selling or advertising products, making their highly suspect. In this category:
Mehmet Oz: https://en.wikipedia.org/wiki/Mehmet_Oz Steven Gundry: https://en.wikipedia.org/wiki/Steven_Gundry
The processing instructions that are well established will also be placed in this project. Ideally they are in a very short document so people can read it, but also could be in an expert system. Thus for example, someone making oxalate stones could get a recipe on making a lentil soup and would be told that soaking lentils or beans will reduce the oxalate count. This is extremely well documented.
Another claim made by Dr. Fuhrman is that eating nuts and seeds containing oils in the same meal as eating greens and berries rich in oil-soluble vitamins increases nutrient uptake. This seems to make intuitive sense, but has to be vetted. Dr. Fuhrman also makes claims that taking a multivitamin containing folic acid is not beneficial, that folate should be obtained from nutritious food. This kind of claim must be carefully vetted as it does not appear to have the same kind of unimpeachable large scale studies behind it.blackberryblackberry