R-bios6643-L16_completed.Rmd

---
title: "BIOS6643. L16 Introduction to Generalized Additive Models (GAMs)"
cheauthor: "EJC"
-- date: "" 
header-includes:
   - \usepackage{amsmath}
   - \usepackage{float}
output: pdf_document
---

\newcommand{\bi}{\begin{itemize}}
\newcommand{\ei}{\end{itemize}}
\newcommand{\itt}{\item}


```{r setup, include=FALSE, echo=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

<!-- -->


```{r installp, echo = FALSE, eval=T, include=F}
##library(dplyr)
##library(tidyverse)
##library(ggplot2)
##library(geepack)
##library(lme4)

library(HRW) ## to load spinal bone mineral data
library(mgcv)


```

# Modeling non-linearity in longitudinal data

## Colorado CF FEV data
The data set gives characteristics of children patients with a diagnosis of Cystic Fibrosis (CF) who are patients at the Colorado Children’s Hospital. 

VARIABLE DESCRIPTIONS:  
id: patient ID	
race: race	
age: Age in years	
gender	: gender
fev: \% of predicted forced expiratory volume in 1 second


```{r eval=T, echo = T, include=TRUE}
#  Read in the data 

##  Read in the data 
dat <- read.csv("/Users/juarezce/Documents/OneDrive - The University of Colorado Denver/BIOS6643/BIOS6643_Notes/data/fev.csv",
                header=TRUE)
length(unique(dat$id)) ## number of individuals
head(dat,3)
summary(dat$age) ## age in years
summary(dat$fev)
dat$sexf <- ifelse(dat$gender=='f', 1, 0)
table(dat$gender, dat$sex)
```

## Smooth effect of age
```{r eval=T, echo = T, include=T}

fit <- gamm(fev ~ sexf + s(age) ,
            random = list(id = ~1), method='REML',
            data = dat)
##attributes(fit)
summary(fit$gam)
plot(fit$gam, shade=TRUE, shade.col='palegreen', bty='l')

gam.check(fit$gam)
summary(fit$lme)


```

## Smooth effect of age by gender/sex
```{r eval=T, echo = T, include=T}
fit.fem <- gamm(fev ~ s(age) ,
            random = list(id = ~1), method='REML',
            data = subset(dat, sexf==1))

fit.ma <- gamm(fev ~ s(age) ,
            random = list(id = ~1), method='REML',
            data = subset(dat, sexf==0))

summary(fit.fem$gam)
summary(fit.ma$gam)
plot(fit.fem$gam, shade=TRUE, shade.col='palegreen', bty='l')
plot(fit.ma$gam, shade=TRUE, shade.col='blue', bty='l')

```


## GAMM including gender/sex using "by" option
```{r eval=F, echo = T, include=F}
fit2 <- gamm(fev ~  sexf + s(age, by=sexf) ,
            random = list(id = ~1), method='REML',
            data = dat)

summary(fit2$gam)
plot(fit2$gam, shade=TRUE, shade.col='palegreen', bty='l')

```

# Weight loss trial (PI: Bessesen) 
## Weight loss trial (Saxon, 2019 J Gen Intern Med) 
- Randomized trial at the subject level to compare a toolboox intervention versus usual care in the primary care setting

- The toolbox consisted of: partial meal replacement program, Weight Watchers vouchers, recreation center membership, phentermine-topiramate ER, phentermine, or a group behavioral weight loss program (Colorado Weigh). 

- 305 individuals were randomly selected to be offered intervention, and 2640 were eligible comparators in the usual care group

- 119 individuals had a baseline visit (305-119=186 did not consent or did not attend baseline visit)

```{r eval=T, echo = T, include=T}

dat.wt <- read.csv("/Users/juarezce/Documents/OneDrive - The University of Colorado Denver/BIOS6643/BIOS6643_Notes/data/weight.csv", header=TRUE)
head(dat.wt, 2)

```

```{r eval=T, echo = T, include=T}

fit.wt <- gamm(wtkg ~  studygrp + s(timemo) ,
            random = list(studyid = ~1), method='REML',
            data = dat.wt)


summary(fit.wt$gam)
summary(fit.wt$lme)
plot(fit.wt$gam, shade=TRUE, shade.col='palegreen', bty='l')
plot(fit.wt$lme)

gam.check(fit.wt$gam)

```




# Notes

- Package gamm4 works similarly as gamm except it uses lme4:lmer instead of nlme:lme as the engine to fit LMM models.