Those of you who have been following my blog only since my move to engineer.diet may be under the impression that my only focus is on data engineering and that my upcoming book will have the same focus. In fact the only reason I am doing data engineering at all is to try and set safe boundaries for what is the core of my approach to diet and workout: control theory.
Basically my approach to diet and workout builds on two ideas:
- Use a control-feedback loop to achieve the best achievable values for solid health control variables.
- Achieve 1 in a way that avoids potentially dangerous input variable levels.
Recently I’ve been putting quite a lot of focus on #2. That is, I’ve been looking into input variables and input variable levels that might, according to all cause mortality figures end up doing more harm than good. In doing so we have identified that low-carb theory, while not being infallible in perfect in explaining our data engineering findings is pretty close to the mark, especially where older age groups are concerned. We also identified the importance of age and gender dependent risk factors. Basically the whole #2 part of my work is concerned with what might kill you. More specifically, with what might kill you despite of #1 markers showing great progress. In this post however I want to revisit #1, that is , I want to focus on whatever doesn’t kill you and helps in achieving a better health.
Markers and controls
Nutritional and medical science for a great part are about markers. For example blood serum markers are used when screening for diseases like cancer. The use of multiple independent markers that correlate with a condition is extremely useful for diagnosis. The condition itself may not be easily checked for, the markers, up to a point can stand in for the condition we are looking for. Risk factors have markers to, and while for diseases like cancer its obvious that trying to use a marker as a way to cure the disease by turning it into a control is futile, when talking about risk factors this may not always be the case. In control feedback theory, a control is an output that is used in a feedback loop to control input variables. The problem with this is that not all risk markers are good risk markers and not all good risk markers end up being suitable as control. Remember that markers are often found based on correlations and correlations may or may not have a causal connection of a useful size in the direction required for an effective control.
Health and exercise
We shall be looking at two markers for general health to be used in a control feedback loop. The first one is one that is often overlooked as provider of useful controls: exercise. It is well established that physical exercise contributes to a better health. Conversely however controls based on exercise that correlate the most with physical health conflict with markers commonly used for an other health related variable.
Obesity is a major health issue and a risk factor with respect to many diseases such as heart disease and diabetes. As such, no control feedback loop should be without an obesity geared component. But to do that we need to look at finding a proper control yet. Commonly used markers related to obesity have proven very poor controls that have led to people prematurely abandoning a healthy exercise regime and that have pushed people towards the destructive dieting regimes that in potential do more harm than good while making people suffer in hunger, loosing weight, but not actually doing much for the overall bodily health. So before we pick our control we shall look at the alternatives first.
What is the unhealthy part of obesity? Is it the weight? Well that is what most of the world seems to want to convince us of. Looking at recent research however, it turns out that there are two components to weight related risk and only one has to do with obesity:
- Bodily fat
- Steroids usage
If we ignore the steroids usage end look at bodily fat, there absolutely is a correlation. But looking at the correlation it is clear that it leaves quite a bit to be desired for weight and BMI to be used as stand in for bodily fat levels. If we add to this that exercise on a healthy diet leads to muscle growth, we must come to the conclusion that while BMI is already a bad marker, it’s even a worse control.
Better controls for obesity.
When looking at the data of obesity related markers and obesity related risks, there are three distinct markers that are more suitable as control than body weight and BMI.
- Total body fat percentage (TBFP)
- Visceral fat levels (VFP)
- Waist-Hip Ratio (WHR)
Visceral fat levels are actually the correlated the strongest with obesity related health issues, but TBFP and WHR are closely tied for second, both leaving BMI far behind in usability as control. We shall be working on all three of these markers in different ways, using one directly as basis for control in our control feedback loop. As we already established our nutritional plan should probably be pretty close to a low-carb diet, and carbohydrate intake has been found to be the main driving force behind a visceral dominance of fat distribution, we consider that one covered by static dietary choices. By picking strength sports, including the essential squatting exercise we cover the hip aspect of the WHR. So that leaves us with the TBFP for use in a control for our feedback loop.
An alternative Body Fat Index.
Depending on age and gender there is a recommended healthy range for your TBFP. Using a simple formula we can calculate a body fat index usable as part of a multi dimensional control. We first define three variables:
- AGE : Your age in years
- GENDER : 0 if you are a female; 1 if you are a male
- TBFP : Your total body fat percentage as measured by a body composition measurement scale as found in most gyms today.
- LBFP: Lower Body Fate Percentage. This is the lowest we should allow our body fat to go without risking underfat levels.
Now the first thing we need to do is calculate our LBFP based on our age and gender:
LBFP=20.5 – 14.5 x GENDER + AGE x (1+GENDER) / 20
Now, amusing that our LBFP is lower than our TBFP, we can calculate our Body Fat Index using the following simple formula:
BFI = (TBFP – LBFP) / 5
This will yield a useful gender, weight and age independent index that indicates how close you are to being the leanest you achievable. But remember, this is just one half of the picture, you don’t want to be just lean, you want to be healthy, and not being obese is just one half of the equation.
A weight relative Body Strength Index.
Next to not being obese, a healthy body is relatively muscular and strong. If you are under eating, your strength going down significantly is a first sign you are not on the right track to a healthier you, but might be messing up your body by a multitude of factors that might leave you thin but unhealthy. So how do we avoid this? We add a second index to the menu, the Body Strength Index. As with the BFI we want the index to be gender and age independent. To do this we look at your relative strength , relative to your body weight. You can be a 40 kg woman and have a better BSI than a 120 kg guy, basically. So lets look at our variables:
- WEIGHT: Your total body weight.
- SQUAT: The average top of your last four squatting sessions.
- BENCH: The average top of your last four benching sessions.
- LIFT: The average top of your last four dead lifting sessions.
- GENDER: Again 0 if you are a female and 1 if you are a male.
Given that you are not in it to win the Olympics, and given that you are working at two connected variables at once that will keep you from reaching extreme strength levels, we set our strongest achievable strength level to a number of times your body weight. A number that is considered quite strong for your particular gender. At that level your BSI should be zero.For a man, we set our ultimate strength goal to 6 times your total body weight. If you are a woman, 4.5 times your total body weight would already be quite impressive.
So here is the Body Strength Index formula:
BSI = 6.75 +2.25 x GENDER – 1.5 x (SQUAT + BENCH + LIFT) / WEIGHT
If the BSI yields a negative number, that means you are pretty damn strong. Use zero for your BSI in that case.
Combining the two: Generic Body Health Index; an alternative to the BMI.
So now we have two indices to replace the old BMI as control, but how do we combine them and how do we plot a path to self improvement? Well to combine them we simply use a single complex number to express our Body Health Index with:
BHI = BFI + i BSI
If you wish to play around a bit with your stats to see where you are at, I’ve put the above is a little Google Docs spreadsheet here.(Please don’t vandalize the sheet and mess it up for others)
Now the key to plotting the ideal path to a healthier you lies in finding a balance between improving the size of the vector and in improving the balance between the two components of the vector. One way we can do this is by drawing a circle fragment with the following properties:
- It passes through our BHI
- It passes through zero at an angle of exactly 45°
We can now draw an arrow defining the initial target direction for our control-feedback loop that follows the direction of the circle at our current BHI. That arrow is our ideal path for now. The difference in heading between the arrow and our actual progress after our next 12 workout sessions will be our error that we shall try to steer on with our inputs. Making this process into a critically dampened control-feedback loop leading you to a balanced BHI as close to zero as humanly possible is the ultimate goal of my approach. One variable at a time. Some variables will lead you astray for a period but in the end you should have a diet and workout plan that is perfectly tailored for your body.
This, more even than the data-engineering part is what should be considered the core of my engineers approach to diet and workout. I hope all of this is making at least some sense. My (projected to be a free e-book) book will include info on how to build and tune a control feedback loop around these ideas.