Projection of US adult obesity trends based on individual BMI trajectories

1. Data

Data from the National Health and Nutrition Examination Survey (NHANES) are freely available at [https://www.cdc.gov/nchs/nhanes/](https://www.cdc.gov/nchs/nhanes/). Details on the variables used here may be found in script `R_preparation/variables_description.R`.

2. Preparation of data (`R_preparation`)

The data for functional data analysis is prepared executing `main_prep.R`, in the `R_preparation` folder. The MATLAB code (folder `matlab/nhanes/`)  is called from within R.

The source code of the **BFDA toolbox** is modified only so as to save to csv files the posterior sample of individual trajectories. 

NB: Dependencies of the BFDA toolbox (e.g. mcmcdiag) are **not** included in this repository. They can all be downloaded from [BFDA's homepage](https://github.com/yjingj/BFDA).

## Analysis of results (`R_analysis`)

Analysis of results of BFDA (e.g. computation of prevalence obese or of time spent obese) is conducted in the `R_analysis` folder. Start from the `main.R` script.

4. `bmi.utils`

A minimal R package that groups functions used for the preparation and analysis steps. 


5. List of scripts in R_preparation and R_analysis

R_preparation folder

main_prep.R : main file for the preparation of data
regions_comparison.R: draw graphs showing on which regions of the CohortxAge plane measured and reported past BMIs can be compared 
graph_recall1y10y.R: compare reported past BMIs (1yr and 10yrs before survey) to measured BMIs
graph_recall25.R: compare reported past BMIs (BMI at age 25) to measured BMIs
variables_description.R: description of NHANES variables used
descriptive_stat folder
    change_smoking.R: draw changing smoking prevalence at age 25 by sex and birth cohort
    change_ethnicity.R: draw changing ethnic composition by sex and birth cohort
    change_education.R: draw changing educational attainments by sex and birth cohort
    plot_bmi_traj_f.R: function to plot individual bmi trajectories 
    plot_bmi_traj.R: effectively draw trajectories
    smoking_continuation.R: plot association between smoking status at age 25 and current smoking status for those with 'completed' smoking history
load_other_nhanes
    nhanesII.R: load data from NHANES III (1976-1980) to assess bias specific to reported past BMIs
    nhanesIII.R: load data from NHANES III (1988-1994) to assess bias specific to reported past BMIs

R_analysis folder

main_analysis.R: main script for the analysis of FDA results
prevalence.R: compute prevalence of obesity and severe obesity at age 55 by cohort
prevalence_race.R: compute prevalence of obesity  at age 55 by race and cohort
prevalence_young.R: compute prevalence of obesity and severe obesity by cohort at age 35
tso.R: compute time spent being obese
tso_obese.R: compute time spent being obese by those obese at age 55

R version: 3.6.0