Session timetable

Day 1: Epidemiological delays

Wednesday - Half day starting at 1:30pm

Session 0: Introduction and course overview

Wednesday: 13.30-14.00

Session 1: Delay distributions

Wednesday: 14.00-14.45

  • Introduction: epidemiological delays and how to represent them with probability distributions (10 mins)
  • Practice: simulate and estimate epidemiological delays (30 mins)
  • Wrap up (5 mins)

Session 2: Biases in delay distributions

Wednesday: 14.45-15.30 and 16.00-16.45

  • Introduction: biases in delay distributions (15 mins)
  • Practice: interval censoring - estimating delays from discrete dates (30 mins)
  • Practice: right truncation - estimating delays in real-time with incomplete data (30 mins)
  • Wrap up (5 mins)

Session 3: Using delay distributions to model the data generating process of an epidemic

Wednesday: 16.45-17.30

  • Introduction: Using delay distributions to model the data generating process of an epidemic (10 mins)
  • Practice: implementing a convolution model and identifying potential problems (30 mins)
  • Wrap up (5 mins)

Day 2: An introduction to nowcasting and forecasting

Thursday - Full day

Thursday: 09.00-09.15

  • Day 1 review (15 mins)

Session 4: \(R_t\) estimation and the renewal equation

Thursday: 09.15-10.00

  • Introduction: the time-varying reproduction number (10 mins)
  • Practice: using the renewal equation to estimate R (30 mins)
  • Wrap up (5 mins)

Session 5: Nowcasting concepts

Thursday: 10.00-10.30

  • Introduction: nowcasting as a right-truncation problem (10 mins)
  • Practice: simulating and basic nowcasting (20 mins)

Session 6: Nowcasting with an unknown reporting delay

Thursday: 11.00-12.00

  • Introduction: joint estimation of delays and nowcasts (10 mins)
  • Practice:
    • joint estimation of delays and nowcasts (20 mins)
    • joint estimation of delays, nowcasts and reproduction numbers (20 mins)
  • Wrap up (10 mins)

Thursday: 12.00-13.00

  • Lunch

Session 7: Forecasting concepts

Thursday: 13.00-13.30

  • Introduction: forecasting as an epidemiological problem (10 mins)
  • Practice: basic forecasting with ARIMA models (20 mins)

Session 8: Improving forecasting models

Thursday: 13.30-14.15

  • Introduction: ARIMA concepts and transformations (10 mins)
  • Practice: ACF plots, transformations, and seasonality (30 mins)
  • Wrap up (5 mins)

Session 9: Forecast evaluation

Thursday: 14.15-15.00

  • Introduction: scoring rules and metrics (10 mins)
  • Practice: evaluating forecasts with various metrics (30 mins)
  • Wrap up (5 mins)

Session 10: Forecasting with ensembles

Thursday: 15.30-17.00

  • Introduction: strategies for collating and combining models (10 mins)
  • Practice: evaluating methods for ensemble forecasts (75 mins)
  • Wrap up (5 mins)

Day 3: Forecast hubs, advanced topics and course summary

Friday - Full day

Friday: 09.00-09.15

  • Day 2 review (15 mins)

Session 11: Evaluating real-world outbreak forecasts

Friday: 09.15-10.30

  • Introduction: hub concepts and data structures (10 mins)
  • Practice:
    • evaluation example with existing hub (30 mins)
    • ensembles with existing hub (30 mins)
  • Wrap up (5 mins)

Session 12: Local hub playground

Friday: 11.00-12.00

  • Introduction: Why a local hub? (10 mins)
  • Practice: getting stuck in with a local hub (45 mins)
  • Wrap up (5 mins)

Friday: 12.00-13.00

  • Lunch

Session 13: Combining nowcasting and forecasting

Friday: 13.00-14.00

  • Introduction: Why combine approaches? (10 mins)
  • Practice: connecting nowcasting and forecasting (45 mins)
  • Wrap up (5 mins)

Session 14: Catch-Up and Deeper Exploration

Friday: 14.00-15.00

  • Catch-up and deeper exploration
  • Wrap up (5 mins)

Session 15: Research talks and wrap-up

Friday: 15.30-17.00

  • Research talk 1: Nick (15 mins)
  • Research talk: Why am I so late?
  • Research talk 3: Nick (15 mins)
  • Research talk 4: Nick (15 mins)
  • Summary of the course (10 mins)
  • Final discussion and closing (20 mins)