deck_02.Rmd

---
title: Introduction to <grn>mrgsolve</grn><BR>Data Sets
author: Kyle T. Baron
date: Metrum Research Group, LLC
output: 
  slidy_presentation:
    number_sections: true
    css: [styles/slidystyles.css,styles/colors.css]
  html_document:
    theme: united
    number_sections: true
    toc: true
    css: [styles/htmlstyles.css,styles/colors.css]

---



```{r, include = FALSE}
library(mrgsolve)
library(dplyr)
library(knitr)
library(lattice)
opts_chunk$set(comment = '.', fig.height = 5, fig.width = 9)
```

```{r setup, echo = FALSE, message = FALSE}
mod <- mread_cache("pk1", modlib()) %>% 
  update(end = 192, delta = 0.2) %>% Req(CP)
data(exidata)
data <- filter(exidata, ID <=10)
set.seed(1222)
```

```{r, echo = FALSE}
pop <- exidata
.out <- 
  mod %>% 
  ev(amt = 100, ii = 24, addl = 3) %>% 
  idata_set(pop, ID <=10) %>%
  mrgsim()
```

# Add a population element to the simulation


- <red>model</red> `%>%` <blu>intervention</blu> `%>%` <grn>Go!</grn> `%>%`  <purp>take-a-look</purp> 

> - `model %>% intervention %>%` <orng>population</orng> `%>% Go! %>% take-a-look`


# One population simulation with mrgsolve

<pre>
mod %>% 
  ev(amt = 100, ii = 24, addl = 3) %>% 
  <grn>idata_set(</grn><red>pop</red><grn>)</grn> %>%
  mrgsim() %>% plot()
</pre>

```{r, echo = FALSE}
plot(.out)
```

# `idata_set` takes in individual-level data

- ID  - one per row
- Typically parameters in columns

```{r}
head(pop, n = 3)

length(unique(pop$ID))
```

- This tells mrgsolve to simulate `r length(unique(pop$ID))` units or individuals

# <red>pop</red> and <purp>mod</purp> are connected via <blu>parameters</blu>

```{r}
head(pop, n = 3)
```

```{r}
param(mod)
```

# What else can we do with `idata`?

Batches of simulations or sensitivity analyses

```{r}
idata <- expand.idata(CL = seq(0.5, 1.5, 0.25))
idata
```


# Note: this is the <grn>event</grn> `+` <blu>idata_set</blu> configuration


```{r}
mod %>% 
  idata_set(idata) %>% 
  ev(amt = 100, ii = 24, addl = 2) %>% 
  mrgsim(end = 120) %>% plot(log(CP) ~ .)
```


# `data_set` is the dosing equivalent to `idata_set`


```{r}
data <- expand.ev(amt = c(100, 300, 1000), ii = 24, addl = 3)

head(data)
```

# Note: this is `data_set` configuration

```{r}
mod %>% 
  data_set(data) %>% 
  mrgsim(end = 120) %>% plot(log(CP) ~ .)
```



# <grn>data_set</grn> can also carry parameters

```{r}
data <- expand.ev(amt = c(100,300,1000), 
                  ii = 24, addl = 3, 
                  CL = seq(0.5,1.5, 0.5)) 

head(data)
```

---
```{r}
data <- mutate(data, dose = amt)

mod %>% 
  data_set(data) %>% 
  mrgsim(carry.out = "dose", end = 120) %>% 
  plot(log(CP)~time|factor(dose), group = ID, scales = "same")
```


# <blk>carry_out</blk> or <blk>carry.out</blk>

Copy from the input data to the output data

```{r}
mod %>% carry_out(dose) %>% data_set(data, dose==300) %>% mrgsim()
```


# What <grn>data_set</grn> and <blu>ev</blu>

```{r}
head(data, n = 3)
```

```{r}
ev(amt = 100, ii = 24, addl = 3)
```

# We have now seen 3 simulation setups

> - `mod + ev = ?`
> - `mod + idata_set + ev = ?`
> - `mod + data_set = ?`

# Wait a minute ...
```{r}
head(data)
```

# Generating data sets

- `expand.ev`
- `ev_rep`
- `as_data_set`
- `ev_assign`


# `expand.ev`

Like `expand.grid`

```{r}
expand.ev(ID = 1:2, amt = c(100,200))
```


# `ev_rep`

```{r}
e <- ev(amt = 100, ii = 12, addl = 14)
ev_rep(e, id = seq(5))
```

```{r, eval = FALSE}
e <- ev(amt = 100, ii = 12, addl = 14, ID = seq(5))
```


# `ev_rep`

```{r}
e <- seq(ev(amt = 100), wait = 36, ev(amt = 50, ii = 24, addl = 4))
e
```
```{r}
ev_rep(e, id = seq(2))
```


# `as_data_set`

```{r}
data <- as_data_set(
  ev(amt = 100, ii = 12, addl = 19, ID = 1:2),
  ev(amt = 200, ii = 24, addl = 9,  ID = 1:3), 
  ev(amt = 150, ii = 24, addl = 9,  ID = 1:4)
)
```
```{r, echo = FALSE}
head(data, n = 10)
```


# Updating parameters 1

```{r}
param(mod)
```

```{r, eval = FALSE}
mod <- param(mod, V = 30)
```
```{r, eval = FALSE}
mod %>% param(V = 30) %>% mrgsim()
```


# Updating parameters 2


```{r}
idata <- expand.idata(CL = seq(0.5, 1.5, 0.25))
head(idata)

```

Put the second row into the parameters
```{r, eval = FALSE}
mod <- param(mod, slice(idata,2))
```

# Updating parameters 3

The update here happens as the simulation progresses

```{r, eval = FALSE}
mod %>% idata_set(idata) %>% mrgsim()
```