Epidemiology

Source: https://github.com/aipoch/medical-research-skills

When to Trigger

Activate this skill when the user mentions:

• SIR, SEIR, compartmental models, R0, reproduction number
• Outbreak investigation, contact tracing, epidemic curves
• Incidence, prevalence, mortality rates, case-fatality ratio
• Risk factors, odds ratio, relative risk, hazard ratio
• Cohort studies, case-control studies, cross-sectional surveys
• DAGs (directed acyclic graphs), causal inference, confounding
• Vaccine efficacy, herd immunity, attack rate

Step-by-Step Methodology

1. Define the epidemiological question - Specify the disease/condition, population, time period, and geographic scope. Determine if descriptive, analytic, or modeling approach is needed.
2. Data characterization - Identify data source (surveillance, registry, survey). Assess case definitions (confirmed, probable, suspected). Check completeness and reporting biases.
3. Descriptive epidemiology - Characterize by person (age, sex, demographics), place (geographic distribution, mapping), and time (epidemic curves, secular trends, seasonality).
4. Measure calculation - Compute incidence rate (person-time denominator), prevalence (point or period), attack rate, case-fatality ratio. Report with 95% confidence intervals.
5. Analytic methods - For causal questions: draw a DAG to identify confounders and colliders. Use appropriate regression (logistic for OR, Poisson/negative binomial for rates, Cox for time-to-event). Apply propensity score methods if needed.
6. Disease modeling - Build SIR/SEIR compartmental models. Estimate R0 from early epidemic growth rate or next-generation matrix. Conduct sensitivity analysis on key parameters (transmission rate, recovery rate, latent period).
7. Interpretation and communication - Translate findings into public health actions. Present results with absolute and relative measures. Discuss Hills criteria for causation assessment.

Key Databases and Tools

• WHO Global Health Observatory - International health statistics
• CDC WONDER / MMWR - US disease surveillance data
• Our World in Data - Pandemic and health metrics
• GBD (Global Burden of Disease) - Comprehensive disease burden estimates
• EpiEstim / R0 package - R0 estimation tools
• DAGitty - DAG drawing and analysis

Output Format

• Epidemic curves with proper time axis (onset date, not report date when possible).
• Measures of association as tables: measure, point estimate, 95% CI, p-value.
• Compartmental model diagrams with parameter definitions and values.
• Geographic maps with rates (not raw counts) and appropriate denominators.

Quality Checklist

• Case definition explicitly stated
• Denominators appropriate (person-time for rates, population for prevalence)
• Confidence intervals provided for all estimates
• Confounders identified via DAG and adjusted for
• Selection bias and information bias discussed
• Model assumptions stated and sensitivity analysis performed
• Absolute and relative measures both reported
• Temporal relationship between exposure and outcome verified

Error Handling

• If required inputs are missing, state exactly which fields are missing and request only the minimum additional information.
• If the task goes outside the documented scope, stop instead of guessing or silently widening the assignment.
• If execution fails, report the failure point, summarize what can still be completed safely, and provide a manual fallback.
• Do not fabricate files, citations, data, search results, or execution outcomes.

Input Validation

This skill accepts requests that match the documented purpose of epidemiology and include enough context to complete the workflow safely.

Do not continue the workflow when the request is out of scope, missing a critical input, or would require unsupported assumptions. Instead respond:

epidemiology only handles its documented workflow. Please provide the missing required inputs or switch to a more suitable skill.