Investigative workflow, and each piece on its own terms.

Understanding a system requires many different tools and approaches. The following describes the core methods and models used to analyze emerging epidemics. When exploring the dynamics of an epidemic, the workflow mimics that of any deep investigation into a system, whether you are concerned about internal business processes or climate change, some things stays the same. Build a pipeline to clean, merge and normalize all the data you will need. This is the hardest part and, because it is the hardest part, it is often skipped over. Unfortunately this means that in many AI adoption cycles, businesses will find the interesting data is "difficult to use". missing. In any scientific endeavor, gathering the data is the most important step. The data and its structure must guide theory. Each section presents a self-contained part of an investigative workflow. Breadth lives here. Tag each card investigation or applied so a reader can filter to what they came for.

02investigation Risk-Factor Inference Which place-level factors drive risk — interpretable baselines vs. learned model, mapped by location. 06applied Data Pipeline Reproducible ingestion of messy, multi-source surveillance data — every conclusion traceable to clean input. 07applied Dynamic Transformations Snapshot to process: growth rates, lags, spatio-temporal features that downstream models depend on. 08investigation LSTM Forecasting Does a learned forecaster beat a mechanistic baseline — and where does it break? Rolling-origin validation. 09investigation Agent-Based Simulation Epidemic dynamics that emerge from per-agent behavior across spatial arenas — an intervention / RL testbed.

[Filtering is a wireframe note for now — wire up tag filter buttons later, or keep it as static visual grouping.]