Performing A Binary Search To Find/isolate The Cause Of Health Problems.

March 27, 2012 in Uncategorized

This was an idea I had the other day as I usually try to keep my biohacking to single variable modification but it takes too long.

My background is compsci so I thought this would help others that don’t have a similar background.

Right now I’m taking a number of supplements in the morning.

Usually an hour or two after I take them I feel sick. I don’t know what is causing it.

Removing one at a time would take too long… so I would like to find the solution faster.

The trick is to do a binary search. It requires log(N) comparisons vs N …

For example if you have 10 pills you take every morning, you can find the problem in at most 4 steps (AKA log(10)) vs 10 steps… which is much faster.

4 vs 10 days is WIN.

Basically you first start off by splitting the input set into two groups. Let’s label them 0-9

0 1 2 3 4 5 6 7 8 9

would be the full set. This is what is making you sick.

0 1 2 3 4
5 6 7 8 9

On day 1 you take pills 0-4 … if that makes you sick you split that again.

0 1 2
3 4

You then try 0-2.

If that does NOT make you sick then you try 3-4.

If that DOES make you sick you split that again.

3
4

If 3 makes you sick, you have identified a cause.

However, you still need to test the other batch because you don’t’ want to fall victim to the single cause fallacy.

But it PROBABLY is a single cause so you take 5-9 and find that this doesn’t make you sick.

Now you try one final attempt at ALL the pills WITHOUT 3 and see if that makes you feel ok.

If that works you’ve isolated the problem much faster.