the algorithms.

We're about to diiiiiive in.

This database is exactly that: a database. It compares 200+ birth control methods against one another to suggest the best option based on what is currently in the market. Learn about how it works.

1. 1. Compiling the birth control database.

   This part was a nightmare. We manually scoured the internet, creating a list of all the birth control brands we could find. Spoiler alert: there are a lot of brands. From there, we went to every brand website to gather the active ingredients (if any) and the related dosages, as well as the phased type of the birth control: mono-, bi-, tri-, or quad-.

2. 2. Compiling the progestin database.

   Part 2 was just as manual as part 1, but not nearly as lengthy. Focusing on a key BC variable, progestin, we read hours of clinical information from a list of sources so long it'd rival an encyclopedia. Progestin was key to us because there are several generations (4 to be exact), and various types of progestin within each generation. Each progestin generation, as well as the progestin types, has its own pros, cons, and hormonal activity levels that we stacked like a layered cake.
   

3. 3. Joining the databases.

   From there, we joined the two databases using the progestin type.
   → The progestogenic activity level was pulled directly from the progestin database: it's the multiple of the activity level (1 through 5, determined through research) and the dosage (in mg). This was repeated for the androgenic activity level.
   → Because progestin has estrogenic activity and many birth controls have an actual estrogen component, the activity level of each brand is a sum of these two things: estrogen dosage + (estrogenic activity level of progestin x progestin dosage) = total estrogenic activity level

4. 4. Establishing relativity.

   After confirming the data was normal and there weren't any outliers, we assigned the progestogenic, estrogenic, and androgenic activity levels a score from 1-5 depending on where it fell, or its percentile, within the dataset.
   
   For example, if the progestogenic level was the median value of the dataset, or the 50th percentile, it was assigned to a score of 2.5.

5. 5. Connecting the dots.

   Armed with the knowledge of how symptoms could be managed using hormones, we compared treatment criteria against the relative hormone level of the brand. This allowed us to determine which brands may be better than others at alleviating specific cycle-related and BC-induced symptoms.

And there you have it.