Introduction to Nowcasting Infectious Disease Outbreaks

Meeting Times:

• Monday, July 13, 9:00 AM – 5:00 PM
• Tuesday July 14, 9:00 AM – 5:00 PM
• Wednesday July 15, 9:00 AM – 12:00 PM

Classroom: TBA

Module Summary:
This module introduces the fundamentals of nowcasting, the task of estimating the current state of an outbreak when recent data are incomplete due to reporting delays. Participants will learn to characterise delay distributions from line list data, understand how right truncation biases real-time estimates, and apply correction methods to produce reliable situational awareness. The module covers the statistical foundations of delay estimation, practical approaches to nowcasting case counts, and methods for estimating the time-varying reproduction number. Participants will also learn how these components can be combined in a joint modelling framework that propagates uncertainty appropriately and extends to short-term forecasting.

Prerequisites:
It is expected that course participants have basic knowledge of statistics, mathematics, and infectious disease epidemiology. It is also expected that participants have basic computer software knowledge and some familiarity with the software R.

Module Content:

• Delay distributions: Introduction to epidemiological delays, their biological and administrative origins, and methods for estimating delay distributions from line list data
• Right truncation: Understanding how incomplete reporting biases real-time estimates and why standard delay estimation fails with truncated data
• Discrete convolutions: How delays transform underlying processes into observed data, and the role of convolutions in linking infections to observations
• Reproduction number estimation: The generation interval, renewal equation framework, and methods for estimating the time-varying reproduction number from case data
• Nowcasting methods: Approaches for correcting right-truncated data, from simple adjustment factors to probabilistic models that account for reporting patterns
• Joint modelling: Combining delay estimation, nowcasting, and reproduction number estimation in a unified framework, with connections to short-term forecasting