# Determining a patients biological age?

I am looking for an algorithm that shows the exact coefficients some variables that have on a hospital patient's chronological age versus their biological age.

I understand a lot of the variables that go into determining ones biological age (i.e. Blood pressure, diet, family history of disease, obesity etc.) but not the amount of impact that it has on the actual scoring of their age.

I have found many sites that will calculate it, however I need the actual variable coefficients and influence in order to produce a proper score for thousands of patient records.

Thanks in advance for your input!

-
How is a "biological age" defined? It doesn't sound like it would have a precise definition, but maybe I am ignorant. – Memming Jan 29 '14 at 19:27
Biological age is considered one's "real" age based on how their body is behaving versus someones actual age. – Pete Jan 29 '14 at 19:47
So for example. A 65 year old who still works out, and eats healthy and is of normal weight has a biological age of 55. – Pete Jan 29 '14 at 19:49
There is no rigorous standard for "biological age". It makes no sense. A fit 65 year old might have a heart that, in some tests, performs as well as the "average" 55 year old, but that's just one organ, and one set of tests. – swbarnes2 Jan 29 '14 at 22:55

## 1 Answer

As the definition of biological age is very ambiguous, I propose that to generate an algorithm, you first need to create a quantifiable definition. I submit:

Biological age is a measure indicating what portion of one's calculated life expectancy is already expended, adjusted proportionately for the average life expectancy of their major demographic.

Wow, that's a mouthful! So, let's break it down.

Life expectancy is calculated :

1. Ascertain a person's major demographic (Geographic residence, Race, Gender and Generation).
2. Find the average life span of that demographic.
• This is negligibly speculative as a living person's Generation has not yet fully perished
• Statistical outliers, such as infant mortality, are generally excluded from this calculation
3. Adjust for known major factors that have accepted statistical impact.
eg:
• Lifestyle
• Current medical conditions
• medical history

Average Life expectancy of subject's demographic was calculated in #2 above.

So, now let's throw in some sample numbers.

Statistical Facts: (totally made up)

• Caucasian males living in the France, born in the 1970s have an average life span of 72 years.
• Exercising 30 minutes or more daily increases life expectancy 6 years
• Smoking decreases life expectancy 7 years
• Not smoking increases life expectancy 2 years
Hold on a second?! Wouldn't not smoking already be accounted for in the "Smoking..." section? Well, no, because our demographic sample are all "unknown" so having specific knowledge would statistically change the results in either direction
• Heart Disease decreases life expectancy 8 years
• Devout religious affiliation increases life expectancy 6 years.
• Family history of diabetes decreases life expectancy 2 years.

Two subjects both Caucasian, french males born in 1974 (40 years old chronologically):
Their mom's actually shared a labor and delivery room!

1. Beavis - known: Smoker with heart disease.
2. Butthead - known: Devout Buddhist, very athletic, non smoker with a family history of diabetes

Beavis' life expectancy is 57 (72-7-8)
Butthead's life expectancy is 84 (72+6+2+6-2)

Beavis has expended 70.2% (40/57) of his life expectancy
Butthead has expended 47.6% (40/84) of his life expectancy

Despite sharing a birthday:
Bevis' biological age is 50.5 years (.702*72)
Butthead's biological age is 34.3 years (.476*72)

Note: It's almost 6 hours past my bedtime, so please excuse any stupid mistakes or miscalculations. The general concepts are what is important

-
From the way you phrased your question, it sounds like you are doing this programmatically. As that was not explicitly stated I reserved this suggestion for the comments: I would recommend storing the statistical factors with both true and false adjustment values (like we did smoking above) as, inherently, any factor's inverse will also have an impact. Then when a factor is unknown you just don't apply any adjustment. Also, in the unlikely event where a factor's inverse is negligible or unknown you can just store an adjustment value of zero. – trex005 Jan 30 '14 at 8:49
The problem with calculations like this is that the time-spans (smoking lets you die 7 years earlier) are based on population wide statistics. They include healthy, 95 year old smokers, which smoke 2 packs a day as well as 35 year old smoking lung cancer patients. What you can derive from this are general recommendations, but hardly specific calculations, as they have to include a lot of personal factors which are unknown. – Chris Jan 30 '14 at 10:20
@Chris Agreed wholeheartedly. This flaw is inherent in statistics universally and is why we use non-precise titles like "life expectation". Additionally, using the following statement highlights the accepted variability: "Adjust for known, major factors that have accepted statistical impact." There are literally infinite variables in play, some may have nanoseconds in actual lifespan impact some could eliminate a teenage athlete mid stride. An infinitesimal fraction of these variables are known, a smaller fraction are studied and the minority of those have an accepted statistical impact. – trex005 Jan 30 '14 at 10:53
Thank you for your input @trex005, I have taken your factors and took a adapted approach with my data variables and took some additional factors from the cdc life tables as well as Moran E. Levine from his article: Modeling the Rate of Senescence: "Can Estimated Biological Age Predict Mortality More Accurately Than Chronological Age?" and from L. Galzigna and M. Cecchettin their article "a simple procedure for calculating biological age." – Pete Jan 30 '14 at 19:06