Discrete dynamic survival analysis of epidemic case line lists

Discrete Dynamic Survival Analysis (DDSA) estimates transmission parameters from influenza A(H1N1), Ebola, and measles case line lists using differentiable discrete-time likelihoods.

• Why it matters: Mechanistic dynamic survival analysis often needs PDE schemes to handle non-exponential infectious periods, making implementation model-specific and limiting broader use.
• What they did: DDSA uses difference equations to fit deterministic models directly to calendar-day onset, recovery, and hospitalisation data, with differentiable log-likelihood via forward-mode automatic differentiation and discrete infectious-period options (boxcar or device-of-stages).
• The result: In three historical outbreaks, DDSA’s transmission estimates were close to published analyses where available and it can handle incomplete line lists; synthetic tests show recovery intervals identify the infectious period and infection times plus final size constrain R0 when the at-risk population is known.

AI summary of the abstract · may contain errors · not medical advice · read the paper.