Skip to content

RA · Foundations

Learning epidemic modelling and forecasting

A learning guide to epidemic modelling and forecasting: core ideas, the systems and evaluations to know, the skills, and how expertise is shown.

As of
17 November 2026
Okuma
11 minutes
Kanıt
10 sources, cited in order
Seviye
All levels: a map for new learners and practitioners
Çalışma süresi
15 minutes
NEXUS TWIN · WF-1 · NETWORKSEED ed4f67ecRA-F07.DISCIPLINE.EPIDEMIC-MODELLING-FORECASTINGxyzSCHEMATIC · NETWORKST-01 · 0.806 0.192ST-02 · 0.157 0.148ST-03 · 0.156 0.277
WF-1 · network Schematic wireframe twin (network, health system capacity). An illustration, not data.

What you will be able to do

  • Tell a model, a forecast and a scenario apart.
  • Describe the role of surveillance and epidemic intelligence.
  • Explain why forecasts are compared with a baseline and scored.
  • Read a short-term forecast as a range and plan for it.
  • Name the main systems and evaluations in the field.
Seviye
All levels: a map for new learners and practitioners
Çalışma süresi
15 minutes
Before you start
None: the guide starts from the basic ideas

01What epidemic modelling and forecasting covers

What epidemic modelling and forecasting covers

Epidemic modelling and forecasting is a discipline of the Faculty of Health and Society at Risk Academy. The discipline page covers the use of mathematical and statistical models to describe how disease spreads, to compare options for control, and to forecast what may happen next.

A model is a simplified description of how something works. A forecast is a statement about what will happen, often with a range of uncertainty. A scenario is a statement about what would happen if certain assumptions held. A learner here must keep these three apart, because each answers a different question and each fails in a different way.

02Why models matter to risk work

Why models matter to risk work

Models turn data into choices. Short-term forecasts of the COVID-19 pandemic in the United States served, in the words of a peer-reviewed evaluation, as a visible and important communication channel between the scientific modelling community and both the public and decision makers 2. The same paper says such forecasts inform short-term decisions such as healthcare staffing needs, school closures and the allocation of medical supplies 2.

Models also sit inside institutions. The UK Government describes the Scientific Pandemic Influenza Group on Modelling, Operational sub-group (SPI-M-O) as a sub-group of SAGE that gives expert advice to the government on COVID-19 based on infectious disease modelling and epidemiology 3. The legal base for sharing information across borders is the International Health Regulations (2005), which WHO describes as an instrument of international law legally binding on 197 countries 4.

Because models guide serious decisions, their limits matter as much as their results. The skill of this discipline is to use a model for what it can do and to say plainly what it cannot.

03How the discipline fits the wider Academy

How the discipline fits the wider Academy

Epidemic modelling shares ground with several disciplines. Public health emergencies and epidemics covers the response that the models inform. Health systems and health care resilience covers the beds, staff and supplies the models try to protect. One Health, zoonoses and antimicrobial resistance covers the animal and environmental side of emerging disease.

Method guides in other faculties help too. The guide to catastrophe, exposure and vulnerability modelling and the guide to model risk and validation cover general modelling practice.

The Nexus Consortium also treats the subject. The Global Centre for Risk and Innovation covers systems modelling as a science. The Global Risks Forum covers the governance of health threats. Nexus Standards, whose standards are proposed drafts, groups early warning work. The discipline's courses are listed in the catalogue as a foundations course and a frontier course.

04A suggested learning order in four stages

A suggested learning order in four stages

The Academy describes study at four levels: introductory, intermediate, advanced and integrative. The four stages below follow the same climb, but they are about subjects, not levels.

Stage one: data and surveillance. Start with where the numbers come from. WHO describes surveillance in emergencies as the continuous, systematic collection, analysis and interpretation of health-related data, and says the data enable monitoring of an intervention and guide priority-setting and planning 1. WHO also publishes Disease Outbreak News, which provides information on confirmed and potential acute public health events of concern across all hazards 5.

Stage two: finding signals early. Next, learn how threats are detected. WHO's Epidemic Intelligence from Open Sources initiative uses open-source data for the detection and assessment of public health threats in near real time 6. The WHO Hub for Pandemic and Epidemic Intelligence was established in September 2021 in Berlin as part of the WHO Health Emergencies Programme 7.

Stage three: models and forecasts. Then study the models themselves, starting with simple ones. The Academy course Health Nexus lists among its outcomes the calculation and interpretation of outbreak and surveillance measures, including attack rate, case fatality ratio, growth rate, doubling time and the reproduction number. Move on to forecasts and how they are scored. The COVID-19 Forecast Hub collected real-time forecasts of COVID-19 hospitalisations, cases and deaths from 2020 to 2024 from dozens of modelling teams 8.

Stage four: advice and decisions. Finally, learn how modelling advice reaches decision makers and how independent bodies check preparedness. The Global Preparedness Monitoring Board describes itself as an independent monitoring and accountability body to ensure preparedness for global health crises 9. In Africa, the Africa Centres for Disease Control and Prevention was launched in January 2017 as the African Union's continent-wide public health agency, to help African Member States respond to public health emergencies 10.

The order is a suggestion. A statistician may begin at stage three and work back to the data. A public health officer may begin at stage four.

Fig. 01Four stages of study in epidemic modelling

Four stages of study in epidemic modellingA rising ladder of four steps: data and surveillance, finding signals early, models and forecasts, and advice and decisions, each with the sources a learner meets at that step.01Data and surveillanceWHO surveillance; outbreak news02Finding signals earlyOpen-source intelligence; WHO Hub03Models and forecastsOutbreak measures; forecast hubs04Advice and decisionsMonitoring bodies; public healthagencies
The suggested learning order of this guide, with the bodies whose work anchors each stage 1,6,7,2,8,9,10. The order is a suggestion, not a required curriculum.1,6,7,2,8,9,10Sources: WHO; PNAS; Global Preparedness Monitoring Board; African Union. Figure CC BY 4.0. As of 17 November 2026.
Figure 1 in words

Four steps rise from left to right. Step one, data and surveillance: WHO surveillance in emergencies and Disease Outbreak News. Step two, finding signals early: Epidemic Intelligence from Open Sources and the WHO Hub for Pandemic and Epidemic Intelligence. Step three, models and forecasts: outbreak measures and the COVID-19 Forecast Hub evaluation. Step four, highlighted, advice and decisions: independent monitoring and public health agencies.

05What a forecast evaluation teaches

What a forecast evaluation teaches

The best way to learn how far to trust a forecast is to read an evaluation. The study of the US COVID-19 Forecast Hub is one example 2. Starting in April 2020 the Hub collected tens of millions of specific predictions from more than 90 research groups. A multimodel ensemble that combined predictions from dozens of groups every week gave the most consistently accurate probabilistic forecasts of incident deaths from April 2020 through October 2021 2.

The same evaluation shows variation. The 27 individual models that submitted complete forecasts showed high variability in skill across time, places and forecast horizons 2. Two-thirds of the models were more accurate than a naive baseline model 2. Accuracy fell as models looked further into the future 2.

Three lessons follow. First, combine several models where you can, since an ensemble was the most consistently accurate. Second, compare any model with a simple baseline, so that you know whether it adds anything. Third, state the horizon: a forecast for next week and a forecast for twenty weeks ahead are different things.

06Core ideas every learner should hold

Core ideas every learner should hold

Six ideas carry most of the weight in this discipline.

A model is a tool, not an answer. It encodes assumptions about contact, testing and behaviour, and its output is only as good as those assumptions.

Data have gaps. Surveillance depends on what is detected and reported, and WHO notes that effective surveillance can improve outbreak detection in emergency settings 1.

Forecasts and scenarios differ. A forecast says what is expected. A scenario says what would follow from a stated choice.

Uncertainty must be shown. The Forecast Hub work evaluated probabilistic forecasts, which state a range and not a single number 2.

Accuracy falls with distance. Longer horizons carry larger errors 2.

Governance shapes use. Modelling groups give advice inside institutions, and rules such as the International Health Regulations shape what is shared 3 4.

07Worked example: reading a short-term forecast

Worked example: reading a short-term forecast

Worked example (teaching values): the numbers in this section are invented for teaching and describe no real outbreak. A regional health office receives a one-week forecast of hospital admissions from three modelling teams. Team A expects 120 admissions, team B 150 and team C 200. The office has 160 free beds.

Worked example, continued (teaching values): the office does not pick one team. It looks at the range, 120 to 200, and asks how each team has performed in past weeks against a simple baseline that assumes this week's number repeats. It plans for the high end by asking a neighbouring hospital to hold spare beds.

Worked example, last step (teaching values): a week later, 170 admissions arrive. The office records which team was closest, notes that the baseline would have said 140, and writes down what it learned. The habit is the point: compare, plan for a range, and score the forecast afterwards.

Fig. 02Reading a forecast as a cycle

Reading a forecast as a cycleA four-step cycle for the forecast example: compare forecasts with a baseline, plan for a range, wait for the outcome, and score each forecast, which leads back to comparing forecasts.01KarşılaştırForecasts against a baseline02PlanPrepare for a range03BekleObserve the outcome04SkorRecord what was learnedScore every forecastafterwards
The worked example of this guide as a cycle, with no real data. The four steps are the Academy desk's teaching frame. The evaluation of the US COVID-19 Forecast Hub included naive baseline models and scored forecasts against reported deaths 2.2Sources: PNAS. Teaching example, no real outbreak. Figure CC BY 4.0. As of 17 November 2026.
Figure 2 in words

Four steps sit on a circle. One, compare: set each team's forecast beside a simple baseline. Two, plan: prepare for a range, including the high end. Three, wait: observe the actual outcome. Four, score: record which forecast was closest and what was learned. The centre reads: score every forecast afterwards.

08Canonical systems and sources to know

Canonical systems and sources to know

A learner should know a short list of systems and sources, and what each holds.

The WHO pages on surveillance in emergencies 1, Epidemic Intelligence from Open Sources 6 and the Hub for Pandemic and Epidemic Intelligence 7 describe how data are gathered and used. WHO Disease Outbreak News reports individual events 5. The International Health Regulations (2005) set the legal frame 4. The peer-reviewed evaluation of the US COVID-19 Forecast Hub, published in 2022 in the Proceedings of the National Academy of Sciences, is one example of how forecasts were assessed 2. The page of the SPI-M-O group shows how modelling advice was organised in one country 3.

09Skills that make the knowledge usable

Skills that make the knowledge usable

Reading an epidemic curve. Know what each axis shows, and what reporting delays can do to the latest points. Using basic measures. Calculate and interpret growth rate, doubling time and the reproduction number. Comparing with a baseline. Judge a model by whether it beats a simple alternative. Combining models. Use ranges and ensembles in place of a single line. Writing for decision makers. State what is known, what is assumed and what would change the answer. Handling health data. Use the least personal data a task needs and publish only figures that cannot identify anyone.

10Common mistakes when starting

Common mistakes when starting

Three habits slow new learners. The first is to read a curve without asking about delays. The latest points of any reported series are often incomplete, because cases and deaths are reported after they occur. A fall in the last few days may be a gap in reporting and not a real change.

The second is to trust one model. Models differ in structure, data and assumptions, and the forecast evaluation found that skill varied across models, places and time 2. When several credible models disagree, the disagreement is itself information about uncertainty.

The third is to confuse a good fit with a good forecast. A model can match the past closely and still predict the future poorly. The test is always how well it predicts data it has not seen, which is why forecasts are scored after the event. Keep a record of forecasts you have read and how they turned out, and your judgement will improve.

11Open courses and pathways

Open courses and pathways

Risk Academy lists two courses that support this discipline. On 17 November 2026 the page for Health Nexus (RAN-305) describes a Level L2 intermediate course of 25 learning hours in seven modules, self-paced, with a sign-in to enrol. The page for Uncertainty and Confidence (RAS-404) describes a Level L2 course of 29.5 learning hours in six modules on calibrated probability, confidence language and decisions under deep uncertainty, also with a sign-in to enrol.

The discipline's own foundation and frontier courses are shown in the catalogue, and on 17 November 2026 the discipline page lists its courses as in preparation. Nothing there is open for enrolment yet. Beyond the Academy, one free route is to read a forecast evaluation in full 2 and then to follow the WHO pages on surveillance 1 6.

12How expertise is shown

How expertise is shown

Expertise in this field is shown by work that others can check. The Academy records learning, not credentials. Finishing a course gives a completion record. It is not a certificate, an accreditation or a qualification, and it does not state that a person is competent in a profession.

Outside the Academy, the best evidence is a body of work: a forecast that was published in advance and scored afterwards, a model whose assumptions are written down, or a brief for a decision maker that a reviewer has checked. Learners are encouraged to keep such a record from their first course.

13Emerging areas to watch

Emerging areas to watch

Open-source intelligence. WHO describes its initiative as using open-source data for the detection and assessment of public health threats in near real time 6. Collaborative forecasting hubs. The US hub ran from 2020 to 2024 8, and its evaluation was published in 2022 2. Pandemic intelligence capacity. The WHO Hub was set up to respond to weaknesses in how countries detect, monitor and manage public health threats 7. Independent monitoring. Bodies such as the Preparedness Monitoring Board report on readiness 9. Check each issuer for the current state before relying on this guide.

What this guide does not claim: it is suggested learning, not a curriculum that anyone must follow, and it is not medical or public health advice. It reports figures as their issuers published them and draws no new estimates from them; the only invented numbers are the labelled teaching values. Risk Academy does not certify competence, and a completion record is not a professional qualification. The Academy records no partners, cohorts or learner outcomes for this subject, and this guide implies none.

Sorular

Sorular ve yanıtlar

Where should a complete beginner start in epidemic modelling?

Start with stage one of this guide and read the WHO page on surveillance in emergencies. Then learn the basic outbreak measures before moving on to forecasts.

What is the difference between a forecast and a scenario?

A forecast states what is expected to happen. A scenario states what would happen if certain assumptions or choices held.

Why compare a model with a baseline?

A simple baseline shows whether a model adds anything. In the US forecast evaluation, two-thirds of the models were more accurate than a naive baseline, which means some were not.

Is this guide medical or public health advice?

No. It is a learning guide. For a health decision, speak to a qualified professional or your public health authority.

Do courses in this discipline lead to a qualification?

No. A completion record shows that you finished a course. It is not a certificate or a professional qualification.

Koşullar

Sözlük

Surveillance
The ongoing collection, analysis and interpretation of health data to find and follow disease.
Forecast
A statement about what is expected to happen, ideally with a range of uncertainty.
Senaryo
A statement about what would happen if stated assumptions or choices held.
Topluluk
A forecast that combines the predictions of several models.
Baseline model
A simple model, such as repeating the latest value, used as a point of comparison.
Reproduction number
The average number of new cases caused by one case.

Check your understanding

  1. What is the difference between a forecast and a scenario?

    A forecast says what is expected. A scenario says what would follow if stated assumptions or choices held.

  2. Why does forecast accuracy matter at different horizons?

    Accuracy fell as models looked further ahead, so a long-range forecast carries larger errors than a short one.

  3. Why score a forecast afterwards?

    To learn which models were closest and whether they beat a simple baseline.

Continue with a course

Kaynaklar

Kaynaklar

10 sources, numbered in the order they are first cited
  1. World Health Organization (2026). Surveillance in emergencies. www.who.int. Accessed 2026-11-17.
  2. Proceedings of the National Academy of Sciences (Cramer and others) (2022-04-08). Evaluation of individual and ensemble probabilistic forecasts of COVID-19 mortality in the United States. pmc.ncbi.nlm.nih.gov. Accessed 2026-11-17.
  3. UK Government (2026). Scientific Pandemic Influenza Group on Modelling, Operational sub-group (SPI-M-O). www.gov.uk. Accessed 2026-11-17.
  4. World Health Organization (2026). International Health Regulations. www.who.int. Accessed 2026-11-17.
  5. World Health Organization (2026). Disease Outbreak News. www.who.int. Accessed 2026-11-17.
  6. World Health Organization (2026). Epidemic Intelligence from Open Sources (EIOS). www.who.int. Accessed 2026-11-17.
  7. World Health Organization (2026). WHO Hub for Pandemic and Epidemic Intelligence. pandemichub.who.int. Accessed 2026-11-17.
  8. COVID-19 Forecast Hub (2026). The COVID-19 Forecast Hub. covid19forecasthub.org. Accessed 2026-11-17.
  9. Global Preparedness Monitoring Board (2026). Global Preparedness Monitoring Board. www.gpmb.org. Accessed 2026-11-17.
  10. African Union Commission (2017-02-02). The African Union launches Africa CDC, a Continent-wide Public Health Agency. au.int. Accessed 2026-11-17.

Kayıt

Editorial record

Facts and course states are given as of 17 November 2026. This guide is reviewed at least every 270 days and whenever a cited source publishes a new edition or a course of the subject opens.

Published by the Academy desk of Risk Academy. To report an error or request a correction, write through the Risk Academy contact page; corrections are dated and listed with the page. The Academy's statement on AI use explains how AI assistance is used and reviewed. Text, visuals and data are licensed CC BY 4.0, except where a cited source states other terms.

Cite this article

Academy desk, Risk Academy (2026, November 17). Learning epidemic modelling and forecasting. Risk Academy. https://risksacademy.com/health-society/learning-epidemic-modelling-and-forecasting-guide/

Republish

Republish or translate this article free of charge, in print or online, under CC BY 4.0: keep the byline and Risk Academy, link to the original and note any change. Figures, maps and twins made by the institutions carry the same licence; third-party data keeps its own credit line.

Attribution to paste
<p>Learning epidemic modelling and forecasting — Academy desk, Risk Academy, Risk Academy. First published 17 November 2026 at <a href="https://risksacademy.com/health-society/learning-epidemic-modelling-and-forecasting-guide/">https://risksacademy.com/health-society/learning-epidemic-modelling-and-forecasting-guide/</a>. Republished under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).</p>

Pay

EmailLinkedInBlueskyMastodon