Running the model on a single region | Running the model simultaneously on multiple regions

Quick start | Reporting delays, incubation period and generation time | epinow() | regional_epinow()

Set up | Data | Parameters | Delays: incubation period and reporting delay | Generation time | Initial reproduction number | Running the model | Default options | Reducing the accuracy of the approximate Gaussian Process | Adjusting for future susceptible depletion | Adjusting for truncation of the most recent data | Projecting the reproduction number with the Gaussian Process | Fixed reproduction number | Breakpoints | Weekly random walk | No delays | Non-parametric infection model

Background | Setup | Data | Estimating infections | Estimating secondary scaling and delay | Forecasting secondary outcomes

Model | Priors

Introduction | Set up | Overview of all priors | estimate_infections() priors | estimate_secondary() priors | estimate_truncation() priors | Prior impacts and choice guidance | Reproduction number | Gaussian Process length scale | Gaussian Process magnitude | Random walk for $R_t$ (alternative to GP) | Observation model: overdispersion | Observation model: scaling | For estimate_infections(): | For estimate_secondary(): | Generation time distribution | Delays (incubation and reporting) | Truncation | Model choice in estimate_infections() | Priors for estimate_secondary() | Delay between primary and secondary | Priors for estimate_truncation() | Truncation distribution parameters | Practical workflow for prior specification | Step 1: Start with defaults | Step 2: Identify candidates for modification | Step 3: Modify one prior at a time | Step 4: Check prior predictive distributions | Step 5: Check model convergence | Common pitfalls and recommendations | Pitfall 1: Over-informative priors without justification | Pitfall 2: Ignoring generation time and delays | Pitfall 3: Estimating too many uncertain parameters | Pitfall 4: Wrong time scale for length scale | Pitfall 5: Forgetting the max parameter | References and further reading | Key papers

Data | Set up | Parameters | Delay distributions | Generation intervals | Reporting delays | Truncation | Completeness of reporting | Initial reproduction number | Weighing delay priors | Estimation and forecasting | Forecasting secondary outcomes | Interpretation | Evaluating forecasts with scoringutils

Overview | Definition | Matérn 3/2 covariance kernel (the default) | Squared exponential kernel | Ornstein-Uhlenbeck (Matérn 1/2) kernel | Matérn 5/2 covariance kernel | Hilbert space approximation | Modelling the reproduction number | References

Infection model | Renewal equation model | Initialisation | Infections | Time-varying reproduction number | Beyond the end of the observation period | Adjusting for susceptible population depletion | Non-Mechanistic infection model | Delays and scaling | Observation model | Truncation | References

Introduction | Set up | Simulating censored delay data | Setting priors | Fitting the model | Checking parameter recovery

Overview | Data and notation | Likelihood | Continuous formulation | Truncation | Discrete observation likelihood | Untruncated approximation | Primary event distribution | Delay families and parameterisations | Priors | References

Component overview | Estimation models | Infection model | Secondary model | Truncation / nowcasting | Delay distribution fitting | Model configuration | Reproduction number | Gaussian process | Delay distributions | Observation model | Options summary | Forward simulation and forecasting | Simulation | Forecasting | Supporting utilities | Data preprocessing | Workflow wrappers | Stan backend

Introduction | Architecture | Relationship between models | Where to look next

What delay distributions represent | Specifying delays | Why naive discretisation is biased | How primarycensored corrects this | Composing multiple delays | Truncation | Fitting delay distributions from data | References

Case studies | Public health surveillance | Literature | By package authors | By others

Setup | The pipeline workflow | Assigning age groups | Surveys | Bootstrapping | Demography | Symmetric contact matrices | Contact rates per capita | Splitting contact matrices | Filtering | Participant weights | Temporal aspects and demography | User-defined participant weights | Weight threshold | Numerical example | Get survey data | Weight by day of week | Weight by age | Apply threshold | Plotting | Using ggplot2 | Using R base | References

Installation | Loading the package | Getting started | The bi_model class | Generating a dataset | The libbi class | Fitting a model to data using PMCMC | Analysing an MCMC run | Predictions | Sample observations | Filtering | Plotting | Saving and loading libbi objects | Creating libbi objects from previous runs | Debugging | Related packages | References

Usage | Using contact matrices with socialmixr | References

Definitions | Prerequisites | Example

Installation | Loading the package | Loading the model and generating a synthetic dataset | Adapt the number of particles | Adapt the proposal distribution | Compute DIC | Convert between LibBi times and actual times or dates | Create inference chains

Deterministic SIR model, observations of prevalence | Deterministic SIR model, observations of incidence | Stochastic SIR model (SDE), observations of incidence | Stochastic SIR model (jump), observations of incidence | Example observation data frame