ketogenic ratio calculations, canadian nutrient file

The Canadian Nutrient File is a handy and only slightly error-ridden source of food nutrition information. Many of the world's governments publish similar resources, and the United States publishes so many that I can't actually keep track of which is used for what. These resources are convenient especially for macronutrient composition, but are also sort of inconsistently convenient for lots of micronutrient information, including vitamins, minerals, lipid profile, caffeine content, and water mass.

Using one of these that includes most of the foods one has access to is a great way to start toying with the parameters you care about in formulating a food pool that you're willing to eat from, or even monodiets and duodiets that will meet the nutritional needs you care about. Want to try to hit potato diet levels of potassium with a duodiet you're willing to eat? Great, just add up pairs of foods in whatever ratios and amounts you like (calorie-matching all pairs with a range of ratios is a great way to start), and make sure to use potassium as predicate for the combined diet.

I'll save you the trouble of having to work out for yourself that you want to be able to clamp the total mass of the food you're considering, and maybe also the mass of certain ingredients, or the mass of certain kinds of ingredients. (These days, my preference is to think in terms of elements which are sources of rare nutrients, elements which serve as tolerable staples, and side items, and then to clamp the maximum mass of each of those in turn.) If you don't do that, you'll quickly find yourself wondering about consuming 45kg of clam juice per day. I checked with my local grocery store, because that seems like a hilarious high-effort shitpost to fail to achieve, but they can't actually supply that much.

Because of shoddy coverage of things like amino acid breakdown, the Canadian Nutrient File isn't actually a great resource if you want to, say, reduce your intake of isoleucine, and perhaps also leucine and valine, but macronutrients tend to be pretty solid.

In the interest of encouraging folks to do some of their own experimentation, I present a copy of all of the foods in the Canadian Nutrient File annotated with two ketogenic ratio calculations.

The first, Shaffer's method, is the näive approximation that you can do in your head, and it's just ketogenic factors (fats) divided by antiketogenic factors (carbohydrates and protein) by mass, and if the ratio is 1 or lower, antiketogenic factors dominate, while ratios of 2 or higher are dominated by ketogenic factors. The range of values between 1 and 2 can be roughly thought about as allowing ketogenesis and ketone use, but not especially promoting that state.

Shaffer's method is just: F / (C + P)

The second, Withrow's method, is pretty dated at this point, but seems to have some clinical and research utility. It uses a more complex equation with various coefficients to reflect the relative ketogenicity and antiketogenicity of the various macronutrients, so that fat is strongly ketogenic but only very weakly antiketogenic, and protein is appropriately somewhere more in the middle. With this scale, significant clinical effects are observed around the 1.7 mark, I believe in terms of both keto-adaptation and in terms of the neurological effects of ketones.

Withrow's method is: (0.9 F + 0.46 P) / (C + 0.58 P + 0.1 F)

I prefer Withrow's method, particularly because it captures some of the moderated effect of protein intake a bit better. I've included both because there's a lot of published literature that uses each one, and it can also be interesting simply to see the difference. You might imagine that the points where they are the most different might also represent foods for which individual variations and environmental factors may be likely to dominate.

There are other methods out there in rarer use, and you can imagine how alcohol might be incorporated, but at some point you want to actually start having a more nuanced model of metabolism either in your head or in your codebase. A good starting point is to do what one does in modelling heat that is available for the production and ripening of fruit, which is to look at how much of each day is spent being how much hotter than some threshold. You might think about ketogenic factor hours as the sum of these, and to increase either the time under the influence of a given ketogenic factor, or to increase the ketogenic factor over a shorter period of time. You might also wonder about those with the lowest ketogenic ratios, and whether there are meaningful thresholds there in terms of maintaining carbohydrate burning and glucose dominance without fuel-switching problems.

The great thing about these ratios is that you can compute them both for ingredients (as provided here) and for whole diets. The even greater thing is that you can then also think about what might need to be added. This is pretty standard fare for developing well-formulated ketogenic diets for epilepsy, where one thinks in terms of units of ketogenic and antiketogenic food and ensures that the balance is maintained. You gotta drink your cream even if you don't want to finish your other food, and all that.

You can use a little light algebra (thanks to someone who helped me get the math right on this, which I was going to need like five tries to do on my own) to figure out how much fat you'd need to add, for instance, to take a given food or diet up to 1.7 with Withrow's method.

Given T as the target ketogenic ratio: ((T C) + (((0.58 T) - 0.46) T)  + ((0.1 T) - .9) F)  / (0.9 - (0.1 T))

For instance, given a diet or a food which contains 84.1g of protein, 49.3g of carbohydrates, and 159.9g of fat, Withrow's method gives a ketogenic ratio of around 1.6, which with around 15.5g of extra fat becomes 1.7. I keep meaning to throw together a friendly little calculator that someone could use to obtain this number, but I'm not much of a web person. I'd love to see Cronometer show the ketogenic ratio for foods and diets, and ideally let you set a target and show your fat need from that. It's an easy enough tool to build, but we shouldn't be proliferating food trackers. One of the big ones should decide they care about ketogenic diets and help people make these decisions. It's simple algorithmic additions in exchange for potentially major benefits for health.

If you want to start thinking about ketogenic ratios, my annotated list of foods from the Canadian Nutrient File is available as a Google Sheet. I would not stake someone else's life on the data in the CNF, which has structural, typographic, and content errors that show up pretty obviously if you look at it even a little bit. This is not medical advice, and you should do your own calculations if you want to make medical decisions based on the ketogenic ratios of food.

Remember to drink your clam juice (Mollusks, clam, mixed species, canned, liquid; 0.60) and eat your aged walrus blubber (Game meat, native, walrus, blubber, aged; 6.4). Get in touch if you'd like to explore nutrition and dietary tooling to support the formulation of ketogenic diets programmatically.