The people working in medicine are for the most part, doing a wonderful job. No, I don't think things are perfect.
One recent topic of extended concern, is obesity. Men, women, children, ethnic groups, one can find concerns about obesity in just about any minority (sub-population) you care to look at.
In general, there is a concern when a person doesn't have enough fat in their body, and there is a concern when a person has too much fat in their body. The distribution of fat can also be of some concern.
When someone is disecting a dead body, it is possible for them to measure the fat content of every organ and location in the body. It's time consuming and probably expensive.
If we do this disection procedure on enough members of any given sub-population, we will start to see differences in fat content.
Few people document their own lives so that medical people will have the data they require to tie observed properties (such as percent fat in the abdomen) to the life experienced at the time they die (and can be disected). The people in medical research are always dealing with a lack of information.
Medical people have noticed that too much fat is a bad thing. We can't disect a patient just to find out how much fat they have, and where it is. Well, what can a medical person do? An easy measurement to make, is to measure the mass of the person (most people call this measuring their weight, weight is an artifact of mass observed in a gravitational or inertial reference frame).
We analyse how body mass and other variables are related to the health factors. Someone looks at the analysis, and draws boundaries of various kinds. This partitions the data. Each area corresponds to some "diagnosis". This data is then used by other medical people in the best way they can, for the purpose of helping people live a good life.
Everything up to this point is fine, it is what happens afterward that can get silly. Subjective judgements (do I look fat?) are probably the first place to start. The medical person looks at you for some length of time, and from that look guesses as to how much fat you are carrying. In order to be accurate and precise, they need a lot of practice at this method. No matter how much practice they get, the precision isn't great. Bias across sub populations is also difficult to avoid.
Another way to diagnose obesity is total body mass, or more usually total body mass compared to people of similar height. Being an obective diagnosis, it is easier to spot problems. Bone is noticably denser than other tissue (even though bone itself has a range of densities), so someone with more bone mass gets lumped in with obese people. Someone who is muscular gets lumped in with obese people. Someone who has small bones and very little muscle, is obese but gets lumped in with the average.
Somebody comes up with the bright idea of using Body Mass Index (BMI). BMI is a person's mass (in kg) divided by the square of their height (in m). Just from the units of this function, it probably shouldn't have much value. But, it gets flogged to death. I will agree it is easy to calculate. It may work fairly well a lot of the time, but it can be completely out to lunch as well. If it wasn't possible to estimate body fat in any other way, there would be an excuse to keep using it. People should quit using it.
The idea that our body is composed of fat and non-fat is called the two compartment model. It doesn't say that non-fat has one density and fat has another, it says that for some mass of say bone, there is a certain associated mass of liver, of heart, of kidney, etc. Since all of these associated masses are fixed, we can divide the body as if it was composed of fat and non-fat, both at some assumed density for each.
In the two compartment model, the only information we need to determine the fraction that is assigned to be fat and the fraction that is assigned to be non-fat, is the density of the person: their mass divided by their volume. Getting their mass was (and continues to be) easy, finding their volume wasn't.
The early way around the volume problem, was an alternative measurement of their weight. We weigh them in a low density fluid (called air), and then we weigh them in a higher density fluid (called water). In the low density fluid, there is insignificant error in assuming their lung volume is zero, but for the in-water weighing, we need the patient to void their lungs of air in order to reduce the lung volume to near zero. And then they have to be still long enough to make the weight measurement. This process is scary, and somewhat expensive.
Today we have 2 alternative ways of measuring the volume of a person, and neither is scary: image analysis and gas law.
The person in question is placed near a number of cameras, and a 3D map of the person is created through image analysis. The method is capable of mapping the outside of a person's body to fairly high accuracy, leaving the volume of the mouth/nose/throat/lungs somewhat unknown.
Whether one assumes ideal gas law, or works with a real gas law, we can estimate the volume to high accuracy and precision, accounting for mouth/nose/throat/lungs, by measuring how the pressure of a gas changes as we change the mass of gas at a known temperature.
If we start from the ideal gas law (PV=nRT), we see that the volume is simply related to the pressure, mass of gas and temperature of the gas. In practice, what we want to do is to put the patient inside a non-scary, comfortable container slightly bigger than the patient is. We then close the container. At this point, we do not know the mass of air inside the container (and outside of the mass of the patient). We then either add or remove a known mass of air, and measure how the pressure changes. This will allow us to calculate the volume of the person.
There are lots of other processes in use to divide the body into fat and non-fat: skinfold, and electrical properties are two of the more popular ones. When used in controlled circumstances, they can give reasonable precision. As a relative measure of body fat, they are probably useful. I am not sure they are accurate enough for non-obvious diagnosis.
There are x-ray and nuclear magnetic resonance techniques available, which are capable of mapping the fat distribution, and hence capable of measuring the volume of fat in the body (and probably the mass as well). I can imagine there are techniques based on radioactive tracers that could be used. None of these techniques is amenable to mass production, they are too expensive.
Our eyes are sensitive to varying wavelengths of light. The wavelengths of light we are sensitive to, are largely those that that can be "excited" by chemical reactions important to life. Or, life on this planet.
We typically have 3 kinds of colour sensitive receptors, which generate the signals we use to differentiate colour. In physics, there is a 3 colour model of light, which was designed to reproduce most of how we percieve the world. There are colours which we can percieve, which are not explained by the 3 colour model. Which is something that most descriptions gloss over. My interpretation of this is: "We cannot see brown.".
Some people are missing one of the 3 receptors, and so have some variation on "colour blindness". It appears that some women develop a 4th colour sensor, and so see a larger number of colours than most people do. Apparently some doctors, who happen to be female, who happen to have 4 colour receptors, are better able to diagnose illness. I wonder if multi-spectral analysis is happening here?
The analysis of visible, and near visible (both infra-red and ultraviolet) light is often termed Multi-Spectral Analysis. Looking at the light reflected from an object as a spectrum, instead of some scalar (for example, it's green) can give you insight into the object. It does require knowledge of the spectrum of light used to "illuminate" the object (you can't reflect or absorb what isn't there).
I see someone has found a way to find blood clots based on the colour of light either absorbed or reflected off them. I'm sure a patent application has either been made, or is shortly forthcoming. I'm sorry, this is just another obvious application of physics. How varying frequencies of light are either absorbed or reflected from matter. Yes, this is useful. Sure, we would like you to make money doing it. No, you cannot have a patent (and an economic monopoly) for this obvious application.