[simon.lykkeland]

I'm an electrical engineer, and right now i am working with a system that is supposed to detect the amount of different substances in urine. Including glucose, urea, creatinine, ketons, proteins, leukocytes, nitrite, urobilinogen, hemoglobin and bilirubin.

The way we want do detect the substances is in an external way, no direct contact with the urine. So we're going to use spectroscopy, sending different wavelength through the sample. Then analyze the the wavelengths to figure out the amount.

So what i'm thinking about now is to order solutions of the different substances, and i have decided to try out urea and creatinine first (the amount og these substances is high in urine). But what should i think about when ordering these solutions in order to make the wavelength spectrum look like the spectrum of real urine? Should the solutions consist of only creatinine and water, or do i need any salts with it?
If you have any other tips or important thing i should focus/think about in this project, then yell out.

Im sorry for my english and my limited knowledge of chemistry.
Best regards Simon.

[Hunter2]

I think it will not work, because most of the substance adsorb in ultra violet light below 400 nm down to 190 nm. So you can not figure which is whhich if you have the mixture. To get an result you need specific indicators which react with on of the substances and crate a colour which can be measured.

[simon.lykkeland]

Thank you for the answer.
You mean i can't differ the substances from each other because the area they absorb will overlap?

[Hunter2]

Yes, exactly.

[Irlanur]

I am also a bit sceptical... Why do you want to do this from scratch? What is "new" about your approach? I am not a specialist in analytical chemistry, but I suppose that one would probably tackle this problem in some sort of LC-MS...

Also:

Including glucose, urea, creatinine, ketons, proteins, leukocytes, nitrite, urobilinogen, hemoglobin and bilirubin.

This things are chemically extremely diverse and I doubt that it is possible to determine the amount of each accurately with a single technique.

[simon.lykkeland]

I dont necessarily need information about all of the substances, but if could get readings from one or two, that would be great. I think i could forget about  leukocytes, nitrite, urobilinogen and bilirubin. Because the concentration of them are really small in urine (0.000322g/L to 0.00526g/L for a positive test).

But the concentration og glucose, keton, creatinine, urea and proteins is quite higher. So if we exclude the other substances, because the concentration is so small, then we are left with fewer substances. Wouldn't it be any ways to get a reading that maybe one of the substances has exceeded a value?

[Arkcon]

Sugars are very difficult to quantify by optical methods.  Their UV cutoff is very low, and everything from solvents, salts, even water and the optics themselves are in the way.  So you have to just forget about that.  That's basic.  The situation is similar for ketones.

You can quantify proteins fairly well at 260 nm, but other things may interfere, particularly urea and creatinine.

This hasn't really been done before, because it isn't really possible, to get useful information from a dilute sample. I mean that, to diagnose human health, a robust, precise, accurate system is needed, and your plan is skimming the very edge of what's physically possible.  That said,we don't really know what you need this assay for.

Also, I find it hard to know why you have a non-destructive limitation for a human waste product.  I mean, I just pee and flush it.  I'm going to discard it anyway, if some assay is needed, you can have some, and conduct a myriad of tests on it.  Colormetric assays are available for all of these components, in the form of dipsticks, or other color reaction assays that use an optical method.  So we need some more information from you about why you have to reinvent the wheel, with some added limitations.

[simon.lykkeland]

Thanks for constructive and helpful answers

Let me explain my project in more details. Mainly it is about detecting the color of the urine of emergency patients in the hospital. My supervisor said if it was possible to detect any substances also that would just be a bonus. So right now i'm just checking the possibilities.

The reason it should be non-destructive is because simplicity and that the urine should not be in contact with the air (because of hygiene). My supervisor also said that we should concentrate on methods that didn't include mixing stuff with the urine, but keep it simple.

I know you can collect a sample of the urine and bring it to the lab, and then get really precise measurements. But this project is about taking measurements right by the bed.

[Furanone]

Near Infrared Spectroscopy (NIR) in the 800-2500 nm range has been used successfully to quantify (estimate) such things as sugars, fats, proteins, certain salts/ash content and moisture content in several foods (eg. very common for milk). The absorption bands are very broad and overlapping in the NIR range compared to mid-infrared where typical FTIR instruments measure where you get many sharp peaks that can form a distinct spectral fingerprint. So in order to make sense of the data from the NIR instrument, chemometrics must be applied (multivariate regression or artificial neural networking). This is something that takes a lot of researching but it is quite interesting in trying to find correlations among data that ostensibly appear totally uncorrelatable. You may not find good statistical fits for all ingredients of interest (eg. hemoglobin, bilirubin and leukocytes may get all grouped together with proteins in general due to their nitrogen content).

If interested, look up Richard G. Brereton as he has several good texts on applied chemometrics, with lots of real-world examples. NIR is non destructive, not incredibly expensive and with fibre optics can be measured remotely in a production environment. It is also more sensitive for ingredients in large amounts of water (such as urine) whereas with mid-infrared, that overpowering water absorption bands would render the data not as useful.