---
title: "Internal function flow"
execute: 
  eval: false
vignette: >
  %\VignetteIndexEntry{Internal function flow}
  %\VignetteEngine{quarto::html}
  %\VignetteEncoding{UTF-8}
---

This document describes the flow of functions and objects within the
package, specifically within the main exposed function
`classify_diabetes()`. It shows the data sources and how they enter or
are used in the function as well as how the different internal functions
and logic are connected to each other. A high-level overview of the flow
is shown in the diagram below.

::: content-hidden
In the mermaid diagram below, invisible edges (`~~~`) are used solely
for layout purposes to help position the nodes.
:::

![Flow of functions in, as well as their required input registers, in
the `classify_diabetes()` function used for classifying diabetes status
using the osdc package. Light blue and orange boxes represent filtering
functions (`keep_*` and `drop_*`,
respectively).](images/function-flow.svg)

The sections below are split into functions for keeping and dropping
events as well as functions for determining the final diagnosis date and
eventual classification of type 1 and type 2 diabetes.

For more details and descriptions about the individual steps within the
algorithm, see the documentation on the internal functions:

-   `prepare_lpr2()`
-   `prepare_lpr3()`
-   `keep_pregnancy_dates()`
-   `keep_diabetes_diagnoses()`
-   `add_t1d_diagnoses_cols()`
-   `keep_podiatrist_services()`
-   `keep_hba1c()`
-   `keep_gld_purchases()`
-   `drop_pcos()`
-   `drop_pregnancies()`
-   `add_insulin_purchases_cols()`
-   `keep_two_earliest_events()`
-   `join_inclusions()`
-   `create_inclusion_dates()`
-   `classify_t1d()`