So, AI can do a lot of things. But it needs input data. So I was thinking, do we currently have any technology that could generate metrics that an AI might be able to train on such that it could estimate calories gotten from a meal. Not what was in the food, but what the body actually absorbed.
Obviously it could be used to make a killer diet tracking app. Cause tracking what you eat is the worst part.
And collecting the metric doesn’t have to be “practical” today. This is just more of a thought experiment. So if currently it would require multiple blood tests per day or something, that would still be interesting to me.
You could probably get quite a bit of useful data by doing gas-chromatography mass-spectrometry on someone’s breath and/or sweat.
Of course, you can’t put a GC-MS in an app, but you explicitly stated that “practicality” was not a concern :)
Hm, I don’t know much detail about those thing. My base knowledge is that those tests would help determine how much of what molecules? was in your breath. But how does that connect to what you absorbed from food?
Stool samples can probably generate the best data set.
Hm. Well I did say what was eaten at the top. And I could see stool giving that information. But I clarified in the middle that I meant what was absorbed. If that was what you were thinking, can you elaborate?
It’s part of the equation. If it comes out, then it’s not absorbed.
It could also show if the body is producing things that are not part of the input. A virus or bacteria would hopefully not be part of the food, but it’ll be part of the excrements.
A blood test can also show some of these things, but it doesn’t show the quantity. Even if everything is all right, the basic things like weight of faeces and volume of urine provides valuable data that is not measured from the food alone.
Ok yeah, part of the equation, that fits.