Machine Learning and Ketosis • Hacker News Board

Re: Machine Learning and Ketosis

#51

Post by MichaelGG » Sat, Aug 13, 2016, 2:34 AM UTC

apineda wrote: post #38

I just started my keto diet about 2 weeks ago and I feel great. Did have one rough night of "keto flu" but then I started supplementing with minerals and more water. I'm super low carb right now. Funny thing is to get (about 4 weeks ago) I would get a double big mac or a double quarter pounder but no drink and no fries and that was it till breakfast. That was fun.

Do you supplement with sodium, potassium and magnesium? It might be placebo but I seem to do better when adding sodium (I drink a gram or two via a broth.) Potassium is easy to get from "lite" salt (hard to eat enough otherwise). The amounts required (several grams of sodium, several of K), according to random sites I've looked up, seem rather high for a non-keto dieter. Dunno if they are accurate.

Re: Machine Learning and Ketosis

#52

Post by MichaelGG » Sat, Aug 13, 2016, 2:36 AM UTC

PerfectElement wrote: post #47

tunnuz wrote: post #3

This is all good if your goal is weight loss. However weight loss doesn't necessarily means higher fitness. Glycogen is fundamental if you do sports, and exercise is a major ingredient in getting fit. If I read correctly, exercise was not a big part of your experiment, how would you suggest modifying the experiment to accommodate one's exercising needs? On a side note, I reached a similar conclusion on the role of "c…

Another anecdote. I tried keto on and off for 2 years while trying to achieve my former triathlete performance. I felt good at the beginning; lost some weight and felt more energy, but it quickly declined and I was never able to perform well. Last year I was introduced to books like Finding Ultra, Whole and Proteinaholic, and went on a high carb diet. Today my diet is based on beans, lentils, potatoes and fruits and…

What's your caloric intake, and how do you feel as far as satiety? I don't always enjoy keto-ish food, but fatty stuff is so filling so fast.

Re: Machine Learning and Ketosis

#53

Post by Tarrosion » Sat, Aug 13, 2016, 2:41 AM UTC

thinline wrote: post #49

hoodwink wrote: post #40

As a long-term ketogenic eater (10 years), here are my top simple tips. Unfortunately they were not gleaned using machine learning. 1. Watch your protein. Most people when first going keto will eat too much protein and not enough fat. Protein has an insulinogenic effect when eaten in quantity. Keep protein below 8 oz per meal. Don't be afraid to eat more fat. 2. Avoid cheese. Yes, it's technically low carb, but it re…

Can you confirm that when you say 8 oz you're only referring to the amount of protein as opposed to the total weight of the food delivering the protein (meat, eggs, etc)?

Got to be the opposite. 8oz is ~240g and 240g of protein is ~1000 calories. For most people that's ~1/2 their daily caloric needs. If in one sitting you are getting half your daily calories from protein alone, something is wrong.

Re: Machine Learning and Ketosis

#54

Post by apineda » Sat, Aug 13, 2016, 2:41 AM UTC

MichaelGG wrote: post #51

apineda wrote: post #38

I just started my keto diet about 2 weeks ago and I feel great. Did have one rough night of "keto flu" but then I started supplementing with minerals and more water. I'm super low carb right now. Funny thing is to get (about 4 weeks ago) I would get a double big mac or a double quarter pounder but no drink and no fries and that was it till breakfast. That was fun.

Do you supplement with sodium, potassium and magnesium? It might be placebo but I seem to do better when adding sodium (I drink a gram or two via a broth.) Potassium is easy to get from "lite" salt (hard to eat enough otherwise). The amounts required (several grams of sodium, several of K), according to random sites I've looked up, seem rather high for a non-keto dieter. Dunno if they are accurate.

I bought a magnesium supplement that includes potassium and sodium, albeit the sodium is the easiest to get anyway, just salt things. Apparently avocados are the potassium go to as well. I've looked at the recommended numbers, and quite frankly I'm not tracking enough to know if I'm hitting htem. The more I remind myself of the recommended and the easier it is for me to calculate on the fly, it's just a matter of looking at my bookmarked page, habit to build I guess until I memorize it. For now I just go by how I'm feeling. If I think I'm going to be frustrated I'll get a little more stringent and careful.

Re: Machine Learning and Ketosis

#55

Post by etangent » Sat, Aug 13, 2016, 2:50 AM UTC

One should be careful extrapolating this type of data to other individuals. For example, sleep on its own may not actually be the primary factor in weight-loss --- the weight gain during periods of lack of sleep may occur simply because you have a habit of snacking while staying up late.

That said, awesome work.

Re: Machine Learning and Ketosis

#56

Post by iopq » Sat, Aug 13, 2016, 2:50 AM UTC

Of course you're going to gain weight when you eat carbs. Glycogen contains mostly water and carbs, so if you store glycogen you gain weight, up to 20 lbs from completely depleted.

So let's say you eat low carb and train very hard for a week. Your muscles will be depleted, your liver will be depleted. You see that you lost 10 lbs! Great! But is it? Maybe you lost one pound of fat and 9 pounds of water. Then you eat NOTHING BUT carbs for three days and you gain 20 lbs. Oh no! But actually it's your muscles expanding their capacity to store glycogen (since you trained hard and depleted yourself they will regain even more glycogen than before). You may have not gained ANY FAT AT ALL!

I have had great success dieting with medium fat medium carb diets. I try to keep between 180lbs and 220lbs (I'm 6'4"). I'm saying that this kind of tracking is oversimplifying the issue of fat loss - you need to see that you're losing fat tissue which involves at least caliper measurements.

Re: Machine Learning and Ketosis

#57

Post by gtrubetskoy » Sat, Aug 13, 2016, 2:52 AM UTC

I don't have a weight problem, I learned about ketosis researching natural ways for improving focus and concentration, and somehow came across "bulletproof coffee". (I'm only using BP for lack of a better name and I don't endorse the stuff they sell online).

To my utter surprise, something as stupidly simple as coffee blended with butter with coconut oil (or MCT) first thing in the morning did unbelievable things for mental productivity. After much googling I learned that this is not a fluke, and that there is real science behind it - beta-oxidation, etc, all that good stuff.

What's most incredible to me is that (1) I didn't know about it, I always thought glucose is the only fuel (and I consider myself fairly knowledgeable as far as basic physiology is concerned) and (2) that sugar, especially the industrially produced kind, is about as close to being the root of all evil as it gets and there is nothing wrong with fat at all.

Re: Machine Learning and Ketosis

#58

Post by stickydink » Sat, Aug 13, 2016, 2:52 AM UTC

Semi-related. For those who want to track their weight daily, efficiently, try Fitbit Aria (or one of the cheaper no-brand WiFi scales). Hook it up to Fitbit app, connect that to TrendWeight (https://trendweight.com/), then throw away the Fitbit app.

Whatever your goals are, however serious you are, I find it a great way to keep track. I just stand on this thing once a morning, and forget about it. And now I have a near-daily record, smoothed out with a weekly moving average.

Re: Machine Learning and Ketosis

#59

Post by paulcole » Sat, Aug 13, 2016, 3:00 AM UTC

gtrubetskoy wrote: post #57

I don't have a weight problem, I learned about ketosis researching natural ways for improving focus and concentration, and somehow came across "bulletproof coffee". (I'm only using BP for lack of a better name and I don't endorse the stuff they sell online). To my utter surprise, something as stupidly simple as coffee blended with butter with coconut oil (or MCT) first thing in the morning did unbelievable things for…

> I consider myself fairly knowledgeable as far as basic physiology is concerned

No offense, but after learning you were dead wrong about a pretty fundamental concept, have you reevaluated this belief?

Re: Machine Learning and Ketosis

#60

Post by anamoulous » Sat, Aug 13, 2016, 3:04 AM UTC

stickydink wrote: post #58

Semi-related. For those who want to track their weight daily, efficiently, try Fitbit Aria (or one of the cheaper no-brand WiFi scales). Hook it up to Fitbit app, connect that to TrendWeight ( https://trendweight.com/ ), then throw away the Fitbit app. Whatever your goals are, however serious you are, I find it a great way to keep track. I just stand on this thing once a morning, and forget about it. And now I have a…

Cosign. I have one of the Withings scales and weigh in multiple times a day. I don't fret over fluctuations -- and I might fluctuate 5lbs in a day -- it's the long term trend line that matters and I can see it on the Withings site. Adding more data just helps.