Epidemic Intervention, Identifiability and Uncertainty Study
A completed synthetic-population mathematics study of SEIR dynamics, parameter identifiability, intervention timing, bootstrap uncertainty and global sensitivity.
Software compatibility
The retained release uses Python 3.14.6. The Docker workflow verifies Python 3.12 compatibility with the same experiment structure and tests.
Project definition
Problem statement
A smooth epidemic curve can hide assumptions about mixing, progression, recovery, reporting and intervention mechanisms.
Different rate and reporting combinations can explain one observed curve while implying different conditional trajectories, so identifiability and uncertainty must accompany fitting.
Project objectives
- Implement and verify a mass-balanced SEIRV model with cumulative incidence.
- Generate overdispersed synthetic reported incidence with a declared reporting fraction.
- Estimate selected parameters and diagnose practical identifiability with profiles, correlations and bootstrap refits.
- Compare contact reduction, faster detection and vaccination mechanisms under fixed assumptions.
- Measure timing sensitivity and global input influence with Latin-hypercube sampling and PRCC.
- Retain configurations, result tables, figures and checksums from one reproducible command.
Project structure
Project components
SEIRV model
Integrates susceptible, exposed, infectious, recovered, vaccinated and cumulative-infection states.
Observation model
Maps latent daily infection flow to overdispersed synthetic reported counts.
Inference laboratory
Fits bounded rates and reporting, profiles beta and performs parametric bootstrap refits.
Intervention study
Compares contact, detection and vaccination mechanisms plus intervention timing.
Uncertainty study
Evaluates a six-input Latin-hypercube design and partial rank correlations.
Evidence builder
Writes trajectories, diagnostics, summaries, checksums and fifteen labelled figures.
Methodology
Project workflow
- 01Load configuration
The study reads the synthetic population, rates, observation process, scenarios and fixed seed.
- 02Verify dynamics
The baseline is integrated and checked for population balance, nonnegative states and monotonic cumulative infections.
- 03Generate and fit
Synthetic reports are sampled and selected parameters are fitted under declared bounds.
- 04Diagnose uncertainty
Profiles, correlations, calibration windows and bootstrap estimates expose parameter coupling.
- 05Compare scenarios
Fixed intervention mechanisms and start days are evaluated against identical baseline assumptions.
- 06Export evidence
CSV, JSON, checksums and figures are built from the same production configuration.
Demonstration scenario
The synthetic baseline has a generating R0 of 2.94 and peaks near 167282 infectious people on day 80. The fitted R0 is approximately 3.318 for the released noisy series. Intervention and uncertainty results are retained as conditional mathematical experiments.
Engineering
Tools and method
- Tools
- The project uses Python, NumPy, SciPy, Pandas, Matplotlib, Jupyter for subject analysis, simulation, and results.
- Dynamic system
- Frequency-dependent homogeneous SEIRV equations solved with adaptive DOP853 integration.
- Count observations
- Constant reporting fraction and negative-binomial overdispersion for synthetic daily reports.
- Parameter fitting
- Bounded nonlinear least squares using signed square-root deviance residuals.
- Global analysis
- Latin-hypercube parameter design with PRCC for peak infectious prevalence.
- Verification
- Automated tests, high branch coverage, lint, dependency audit, Docker reproduction and document validation.
Testing
Evaluation
Evaluation measures
- Population-conservation error and cumulative-incidence monotonicity
- Generating values against fitted beta, sigma, gamma, reporting fraction and R0
- Local fitted-parameter correlation and beta-profile shape
- Bootstrap distributions and parameter coupling
- Peak infectious prevalence, peak day and attack rate by scenario
- Intervention start-day sensitivity
- Uncertainty ensemble spread and PRCC input ranking
Project boundaries
- The study uses a generic synthetic population and does not model a named pathogen, place or current outbreak.
- The intervention values are mathematical assumptions, not measured program effectiveness.
- Homogeneous mixing, fixed rates, immediate vaccination protection and constant reporting are explicit simplifications.
- The project does not provide medical advice, forecast a real outbreak or recommend public-health action.
- Real use requires current surveillance, domain experts, ethical review, governance and external validation.
Included
- 01Complete Python source code
- 02SEIRV equations with cumulative incidence and conservation checks
- 03Synthetic negative-binomial reported-incidence generator
- 04Bounded parameter fitting and calibration-window analysis
- 05Beta profile, local correlation and 80-replicate bootstrap
- 06Four intervention scenarios and nine start-day experiments
- 07320-row Latin-hypercube uncertainty and PRCC study
- 08Fifteen reproducible figures and complete machine-readable results
- 0993-page project report in PDF and editable Word formats
- 1012-page setup and usage guide in PDF and editable Word formats
- 1145 annotated references including current 2026 literature and guidance
- 1214 automated tests with above 98 percent branch-aware coverage
Project record
No information is collected on this page.
- Permanent project ID
- GP-MA-0X9O2PO
- Catalogued
- 21 Aug 2026
- Completed
- 28 Aug 2026
- Verified
- 28 Aug 2026
- Demonstration
- Included in repository
Handover
After purchase
- 01Payment is confirmed
The project is marked unavailable and cannot be purchased again.
- 02Repository access is granted
The buyer's submitted GitHub account receives access to the private repository.
- 03The purchase record is delivered
The certification sheet is prepared from the reviewed buyer details and sent privately by email.