---
title: "Health Data: Mixed Types and Disclosure Risk"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Health Data: Mixed Types and Disclosure Risk}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>")
```

# Synthesizing EHR-like Data

Health records mix continuous lab values, categorical diagnoses and
outcomes. This vignette synthesizes such data and audits disclosure risk.

```{r}
library(DPSynth)
set.seed(3)
n <- 400
ehr <- data.frame(
  age        = as.integer(pmin(100, pmax(18, round(rnorm(n, 60, 15))))),
  labs       = rnorm(n, 5, 1.5),
  diagnosis  = factor(sample(c("I10", "E11", "J44", "N18"), n,
                             replace = TRUE)),
  readmitted = factor(sample(c("no", "<30d", ">30d"), n,
                             prob = c(.6, .15, .25), replace = TRUE))
)

res <- dp_synthesize(ehr, method = "copula", epsilon = 2, n_synth = 400)
head(res$synthetic_data)
print(res)

# Empirical disclosure risk beyond the theoretical epsilon guarantee
res$risk$membership$risk_score
res$risk$attribute$mean_disclosure_error
```

Downstream validation -- predicting readmission from synthetic training
data and testing on real held-out records:

```{r}
ds <- evaluate_downstream(res$synthetic_data, ehr, target_var = "age")
ds
```