ACCIDDA develops infectious disease models, combines public-health data, and trains researchers and public-health practitioners. Its work includes forecasting and analysis for smaller populations, particularly rural communities. As the Insight Net coordination center, ACCIDDA organizes network meetings, technical workshops, and modeling challenges.

Our public work on machine learning for infectious disease forecasting includes:

• Influpaint (GitHub, paper, documentation) uses diffusion models and inpainting to generate forecast trajectories.
• LEMMA-forecast combines predictions from mechanistic models with machine-learning adjustments based on their errors.

Public dashboards present surveillance data, forecasts, and vaccination estimates:

• RespiLens (GitHub, visit RespiLens) displays local respiratory disease data and forecasts, including CDC data. It also includes the Forecastle forecasting exercise and an analysis toolbox.
• Forecast-viz Creator generates public forecast dashboards.
• The NC Measles Vaccine Coverage dashboard shows estimated vaccine protection across the state, derived from our imuGAP tool.

We develop calculators and simulation pipelines for comparing interventions and running modeling studies:

• tti provides a calculator to evaluate Test-Trace-Isolate interventions. Learn how to use it here.
• FlepiMoP2, the Flexible Pipeline for Interchangeable Model Processing, runs configuration-defined simulation and analysis campaigns. Its provider interfaces separate model specification, numerical integration, parameters, processing and persistence. See the current documentation. The earlier flepiMoP remains available for legacy workflows.

These public tools support model construction, parameter specification, simulation, and scenario comparisons:

• tabularepimdl, the Tabular Epidemiological Model framework, provides a language for specifying and modifying infectious disease models.
• paramix scales parameters for compartmental models. See the method paper.
• hashPRNG supports matched random-number streams for scenario comparisons.
• op_system compiles PDE/field, nonlocal and reaction model specifications into shaped numerical evaluators and operator contracts, with reaction metadata, named observables and forcing schedules. The core is published on PyPI; see its capabilities and examples.
• op_engine supplies PDE/nonlocal operator methods and deterministic/stochastic reaction solvers, including independent NumPy SSA banks, observations and replay. The core is published on PyPI; its qualification matrix states the tested CPU methods and backends. Both cores can be used directly or through separate FlepiMoP2 provider packages.

op_infer is in development. It connects Bayesian fitting through established inference libraries, simulation-based calibration and replicated study-design checks. It remains private, pre-alpha and unreleased while broader interfaces and application studies develop.

We also develop modeling packages focused on key epidemiological questions:

• imuGAP estimates vaccine coverage by age and place from vaccination data recorded at different levels of temporal, spatial, and age aggregation.
• hestia allows users to build custom disease and observation models to evaluate household study data and estimate key epidemiological parameters.
• SeverityEstimate is an R package for estimating the severity of infections from line list data that include active and passive detection.
• vaxflux is a python package for modeling seasonal vaccine uptake like influenza vaccines.

Please contact project maintainers for specific questions and if you would like to contribute.

For more general inquiries, to be connected with an ACCIDDA member expert on a particular project, or to talk to us about sponsoring infectious disease modeling research and development reach out to ACCIDDA.info@unc.edu