[jankor]

The full study is much more interesting: "We devised a machine-learning algorithm that integrates blood parameters, dietary habits, anthropometrics, physical activity, and gut microbiota measured in this cohort and showed that it accurately predicts personalized postprandial glycemic response to real-life meals."

http://www.cell.com/cell/pdf/S0092-8674(15)01481-6.pdf

[mbubb]

Thank you for the link - the study is very interesting to read. I have been bumbling around for a while trying to see what fits for me and there have been counter-intuitive things that seem to work. This study makes me think of that. Part of diet is finding what works to keep things in balance, cravings and spikes might indicate the diet is not working.

As an example - initially starting dieting, I stayed away from fruit as a vehicle of carbs. Over the past two weeks I have allowed myself bananas, apples, melon and berries as breakfast and afternoon snack. This appears to not 'spike' me as I would have thought but gives me stability.

Night snacking is a killer for me and I have been able to get by with a cup of tea around 2200 with a single biscuit.

Its just been 2 weeks but these two little cheats actually seem to give the diet some stability - for now anyway.

This study is really useful for me.

[ozy123]

What are you using to monitor spikes? Trying to locate something that would give continuous readings rather than manual testing after every meal.

[rdancer]

Would continuous glucose monitoring devices used by diabetics be suitable?

[wdewind]

This is a great study, for nutritional science I am very impressed with the methodology. It's very large and very thorough. It should be heavily noted that the goal of the study is not to create algorithms for healthy people, but for diabetic and pre-diabetic populations.

The only well supported long term consequences of post-prandial high glycemic response in otherwise healthy people is an increased psychological urge to eat. Outside of that, the importance of insulin levels in healthy people is extremely overhyped. These machine learning algorithms are cool, and may be useful for diabetic people, but for the general population monitoring your insulin levels is a waste of time.

[beachstartup]

> Outside of that, the importance of insulin levels in healthy people is extremely overhyped. These machine learning algorithms are cool, and may be useful for diabetic people, but for the general population monitoring your insulin levels is a waste of time.

a waste of time? really? what does that mean? does this mean the difference between 10-20 pounds over the course of 10 years? because that's the vast majority of 'weight' concerns - people who are overweight but not obese are 'otherwise healthy' but if we can put a finger on exactly what causes people to put on a few pounds it would be a significant advance in science that would affect billions of people.

like, how can you possibly say that investigating what causes a "increased psychological urge to eat" is some kind of trivial pursuit? that's the whole problem with "otherwise healthy" people.

[wdewind]

Sorry I think we agree! My point is it is a psychological issue. All things being equal (calories, nutrients etc.) the speed/magnitude of the insulin response is not important for overall health (in the non-diabetic population). The study mentions things like:

> postmeal blood glucose and its long-term metabolic consequences.

and the point is there aren't long term metabolic consequences.

> but if we can put a finger on exactly what causes people to put on a few pounds it would be a significant advance in science that would affect billions of people

I totally agree! We know it's not the insulin response in and of itself that is causing weight gain, it's the subsequent increased caloric intake, so my point is by being deliberate about what we are studying we can get better results. I'm not saying we don't need to study the psychology of eating, definitely the opposite! That's really what we need to focus on, and stop worrying about the "metabolic consequences" of the insulin response.

[beachstartup]

how do you define 'metabolic consequences'? i would consider 'uncontrollaby wanting to eat more' a 'metabolic consequence'.

and how exactly do you know that the insulin response doesn't affect the psychology? that's begging the question, and that's exactly what i'm saying is erroneous about your mindset. the state of nutrition science is currently horseshit, any casual observer can see that - we need all the good science we can get.

[wdewind]

I actually think the insulin response 100% affects psychology (which is why I think we are in agreement here, still). At this level it gets very tricky to differentiate between "just the psychology" and a physical reaction, though, and so my point is there is no evidence of any bad health effects caused directly by the insulin response. An insulin blood spike is not unhealthy for you in and of itself, but sometimes it will cause you to eat too much and that's not good for you. That gets into will power, which is the place where the mental and physical meet, and where both groups of models tend to not work particularly well. It's not well understood and it's Complicated.

I agree with you about the sorry state of nutritional science, and when it comes to diet I'd love to see a lot more resources put into the psychological (especially reward pathway/will power) aspects of it.