Logo Lanfrica

PeterJohnston296/TiNTS-study-analysis

Domaine:

healthcare

Type de record:

software
Créateur:
Pet
Hôte:
Reproducible analysis code for the TiNTS prospective longitudinal household cohort study of enteric non-typhoidal Salmonella in Malawi. # TiNTS study analysis This repository contains the analysis code supporting the TiNTS manuscript: **Frequent asymptomatic carriage and household transmission of non-typhoidal Salmonella in urban Malawi: a prospective longitudinal cohort and modelling study** The repository contains three main components. ## 1. Epidemiological analyses The top-level Quarto documents reproduce the principal R-based epidemiological analyses: 1. `01_variable_preparation.qmd` 2. `02_elastic_net_selection.qmd` 3. `03_cox_models.qmd` 4. `04_episode_burden_and_poisson.qmd` 5. `05_ct_threshold_sensitivity.qmd` These cover preparation of the epidemiological analysis dataset, elastic-net variable selection, first-event and recurrent-event Cox models, episode burden and recurrence analyses, and Ct-threshold sensitivity analyses. The workflows expect the corresponding analytic inputs under: - `data_raw/` - `data_clean/` - `rds/` Generated outputs are written beneath `outputs/`. The public workflows are streamlined from the development analysis notebooks. Exploratory code, superseded analyses, and manuscript-layout code have been omitted while retaining the analysis steps required to fit the reported models and generate the principal numerical outputs. ## 2. Bayesian transmission models The `bayesian_models/` directory contains the hidden Markov models used to estimate latent acquisition, clearance, and household dependence. It includes: - **Model A:** individual susceptible–infected hidden Markov model; - **Model B:** household-coupled hidden Markov model; - the corresponding Stan model files; - R/Quarto workflows documenting data preparation, priors, sampler settings, model fitting, and posterior summaries. Model B is computationally intensive because household latent states are modelled jointly and may require multi-day runtime depending on hardware. See `bayesian_models/README.md` for details. ## 3. Genomics pipeline The `genomics_pipeline/` directory contains the bacteria …