You may not realize it, but your question, as posted, is very nearly impossible. You going to have to way reduce the requirements, if you want the task to be moderately complicated, or moderately complicate the tests, for a reasonable subset of the analytes you want. Analyzing "everything" with a simple spectrometric method is just simply impossible.
Fats and carbohydrates are notable for not having a chromaphore in the UV-Vis, so a spectrophotometric method is plain out, in that case. We may derivitize some subset for visualization, but that's adding a variety of complications. Total protein may be doable, but mammalian milk contains the opaque protein casein, and that will just wreak havoc with optical methods.
Now, IR methods might work, but you will just get a jumble of peaks from related compounds, so that may not be what you want.
Now, chromatographic methods, either HPLC or GC, may be useful. Gas chromatography may be especially useful for fats and fatty acids. And if you can get a separation of a variety of components, on GC or HPLC, then you may generate a "fingerprint" -- a series of peaks that you can use to see changes in the sample over time. Even though, this fingerprint doesn't tell you exactly what's happening, you can use it to infer what phsiological changes you're trying to understand by this analytical assay. And can say where you're going to focus on for further research.
Possibly, the saddest part is that you have a very complex set of analyses, an extremely vague question, and your best research is Wikipedia. You're really like the proverbial absent minded professor, who didn't know how to drive, but was sure they could get the gist by reading the automobile instruction manual. Simply put, when performing an analysis that we hope to have work, we focus on removing the interference of the matrix. Your sample is simply too complex for the best analytical method, and spectrophotmetric assays are too simple even for moderately complex samples.