The top 10 things to do in Punggol for Epidemiology students are to decode NEA’s dengue-cluster definition, compare new weekly cases with cumulative counts, calculate properly denominated population rates using SingStat, distinguish exposure from illness and notification, model sensitivity and specificity, read regional PM2.5 data accurately, investigate sampling bias around parks and libraries, separate association from causation in housing-and-health studies, write an ethical outbreak information brief and draft a study proposal without collecting any patient records.

Best first activity: start at home with NEA’s current dengue cluster map and write down its actual case, distance and time criteria. NEA uses at least two cases with symptom onsets within fourteen days and relevant addresses within 150 metres for an operational cluster. A map outlines a response area; it does not diagnose every person living there. This is a more useful lesson than trying to collect health information from strangers in the neighbourhood.

This October 2026 guide is different from a general Public Health article. It teaches epidemiological measurement: defined populations, denominators, reporting periods, case definitions, uncertainty, bias and responsible inference. The numbers in worked examples are simulations unless an original authority is explicitly cited. No student should publish fabricated health alerts, acquire private health records or describe a model result as a current Punggol outbreak.

Choose an epidemiology skill

• Beginner: learn what counts as a case and why notification and onset dates differ.
• Numeracy: calculate rates with population denominators rather than comparing raw counts alone.
• Statistics: distinguish weekly incidence, cumulative series and test-performance measures.
• Environmental epidemiology: preserve geographic scale and timestamp when using haze or water information.
• Study design: identify selection bias, confounding and the limits of observational association.
• Communication: write a calm accurate summary with source date and no identifying case information.
• Ethics: use published aggregate information or fictional data unless real research has proper authorisation.

The three quantities behind every epidemiology headline

First define the numerator: the events counted, with their case definition and reference period. Second define the denominator: the population at risk, measured in a compatible area and time. Third define what a rate means, including its scale—for example events per ten thousand residents over a stated period. A number may be arithmetically correct but epidemiologically meaningless if residents are compared with workplace-based cases or a count from one month is divided by a population from the wrong geography.

A surveillance map and an individual diagnosis also answer different questions. NEA’s operational dengue cluster criteria are used for targeting vector-control action; they are not proof of the transmission pathway for each case. National weekly case counts help summarise reporting trends but do not identify the exact risk for every block. A learner who keeps these distinctions clear will make more trustworthy graphs and more ethical statements.

Third, every result has uncertainty: not all illnesses are observed or reported simultaneously, tests can produce false positives or negatives, individuals have different exposure histories, and a convenient sample may miss many people. Good researchers explain what data they used and what data they do not have. Precision in writing is as important as precision in arithmetic.

Ten epidemiology learning investigations

Decode NEA’s actual dengue cluster definition

Where: NEA’s live dengue cluster page, studied privately from a Punggol home or library.

Why this is an epidemiology question

NEA defines an operational dengue cluster as a locality with at least two dengue cases whose symptom onsets fall within fourteen days and whose relevant residential or workplace addresses are within 150 metres. These numbers are criteria for public-health intervention, not a mathematical statement that all people inside the outlined zone are infected. Its alert levels distinguish red high-risk clusters with ten or more cases, yellow clusters with fewer than ten, and green clusters under surveillance after new cases stop. Students must read definitions before colouring a map or comparing two neighbourhoods.

How to do the exercise

Read the current NEA page and write down four fields: onset period, proximity rule, relevant location, and cluster status. Make three hypothetical case cards with invented dates and distances, deliberately labelled simulations. Decide which pair meets the stated time and distance thresholds and which fails, without assigning an actual Punggol address. Draw an abstract map with anonymous points and a legend. Explain why a cluster boundary is a public-health operational area rather than a personal diagnosis or proof of cause.

Worked example

Invented scenario: case A has onset on 1 October; case B on 10 October and lives 120 metres from A’s relevant address. Their dates are nine days apart and the distance falls within 150 metres. This is compatible with the basic timing-and-proximity conditions, assuming the stated address context applies. Case C with onset on 30 October falls outside the fourteen-day window relative to A. Whether a real NEA cluster is declared involves official records and investigations, not a student’s hypothetical map.

What to investigate next

Examine what changes when a date is symptom onset rather than the day someone received a laboratory result. A calculation using notification dates may produce the wrong pairings. Ask why addresses alone cannot establish who infected whom or where exposure actually happened. Students should explain why infection chains require specialist epidemiological investigation, uncertainty assessment and other data beyond point geometry.

Ethics and data limits

Never publish identifiable case addresses or presume that a named Punggol block is a current cluster. NEA’s list is live and time sensitive; as of its 21 October 2026 snapshot the displayed active cluster list can change. Do not label neighbours as infected or create fake public alerts.

Official reference: Decode NEA’s actual dengue cluster definition. Verify the current status and reporting date before quoting live information.

Separate new dengue cases from cumulative cases

Where: NEA’s dengue cases and epidemiological-week pages.

Why this is an epidemiology question

NEA explains that day-to-day dengue notifications fluctuate and weekly figures generally better represent trends. A cumulative figure is not the number of new cases in the most recent week. Epidemiological week boundaries need to be explicit because a count for Sunday-to-Saturday is not necessarily directly comparable with a calendar month total. This is one of the most common errors when writers make a chart from public health dashboards.

How to do the exercise

Use NEA’s dengue cases information and record the reporting unit, date range and whether each displayed number is daily, weekly or cumulative. Create an entirely fictional four-week table and derive a cumulative series by adding each week’s new cases once. Label the values ‘practice data’ and do not position a real Punggol marker next to invented numbers. Draw two separate graphs: weekly new cases as columns and cumulative cases as an increasing line. Explain which graph better reveals whether new reported illness has declined.

Worked example

Fictional weeks show 8, 12, 10 and 6 new cases. Their cumulative totals from a zero starting point are 8, 20, 30 and 36. The final cumulative value rises from 30 to 36 while the weekly incidence falls from 10 to 6. Those statements are consistent: adding new cases always raises a cumulative count, even when fewer occur than in the previous week. A student who interprets the cumulative rise as an accelerating outbreak has confused two different quantities.

What to investigate next

Investigate reporting delays, revisions and the difference between onset, notification and publication. What happens when an earlier count is later corrected? A defensible dataset stores its retrieval date and source instead of silently overwriting numbers. Older students can create a simple rolling average but should explain how smoothing may conceal an abrupt rise.

Ethics and data limits

Do not turn national weekly case counts into a Punggol-specific number without an appropriate official local dataset. A student’s graph cannot diagnose a public-health threat, predict a particular person’s disease or replace NEA’s current advice.

Official reference: Separate new dengue cases from cumulative cases. Verify the current status and reporting date before quoting live information.

Learn the incidence-rate denominator from SingStat’s population tables

Where: SingStat 2026 planning-area data for Singapore residents, including Punggol.

Why this is an epidemiology question

The Department of Statistics publishes 2026 geographical distributions by planning area, age group and other characteristics. Its Population Trends 2026 report notes that Punggol, Tengah and Tampines have relatively high proportions of very young residents, but a relative age proportion is not an infection rate. Epidemiologists compare events to an appropriately defined population at risk and a consistent time period. Simply saying one neighbourhood has more cases without accounting for population size can give a misleading impression.

How to do the exercise

Read the SingStat geographic distribution source and distinguish the planning area from subzones or constituencies. For a first exercise, invent two populations and case counts rather than download real sensitive health records. Calculate cases divided by population, multiplied by a stated constant such as 10,000, and label it a crude illustrative rate. Record the period, case definition and residence denominator. Ask whether the denominator includes the same people and interval as the numerator.

Worked example

Fictional districts A and B report 20 and 15 cases. A has 40,000 residents, B has 15,000. Rates per 10,000 are 5 for A and 10 for B, despite A having more cases in absolute terms. Neither district represents an actual Punggol health outcome. A learner who ranks A as automatically higher risk from 20 versus 15 has ignored the population denominator; the corrected comparison is specific to the scenario.

What to investigate next

Ask how age distribution, migration and daytime workplace population might change the appropriate denominator. A rate among residents may not be meaningful for events counted by workplace address. Introduce standardisation conceptually: comparing crude rates across populations with different age structures can obscure important differences. Explain why a classroom rate is not a personal forecast.

Ethics and data limits

Use only properly released aggregated public data; never obtain individual diagnoses or identifiable residential addresses. SingStat figures need their actual reference date and geographic definition. Do not infer disease status from a population age chart.

Official reference: Learn the incidence-rate denominator from SingStat’s population tables. Verify the current status and reporting date before quoting live information.

Understand the difference between a case, infection and exposure

Where: NEA’s public dengue transmission and prevention information.

Why this is an epidemiology question

NEA explains that dengue virus is transmitted by infected Aedes mosquitoes and describes breeding sites and prevention. ‘Exposed to a mosquito’, ‘infected with dengue’, ‘has symptoms’ and ‘notified as a case’ refer to different events. Epidemiology needs an operational case definition before counts make sense. A mosquito bite is not a dengue diagnosis and the existence of a reported cluster does not prove an individual case was acquired at home.

How to do the exercise

Draw a simple conceptual chain beginning with a potentially infected mosquito, possible transmission to a susceptible person, infection, symptoms where applicable, care seeking and notification. Mark where uncertainty enters at every step. Use official public material about transmission without attempting to capture insects or infer symptoms in others. Invent a clearly fictional scenario in which two people see mosquitoes but only one seeks medical assessment for illness, then ask which facts would actually be needed to classify cases.

Worked example

A student says, ‘Five people saw mosquitoes, so five dengue cases occurred.’ Sightings do not establish infection. The corrected record distinguishes five reported encounters from any laboratory or clinically assessed cases. It also notes that a person who feels unwell needs qualified medical evaluation, not classmates diagnosing the cause from the geographic setting.

What to investigate next

Study how a formal surveillance system could specify case classifications and how misclassification affects calculated rates. Ask why under-ascertainment occurs when not all infections or symptoms are captured in the same way. Draw a branching flowchart indicating suspected, evaluated and confirmed statuses without assuming all cases progress through identical steps.

Ethics and data limits

Never attempt to infect, rear or catch mosquitoes for an informal project. Do not ask actual patients for symptoms or private records without appropriate institutional research approval. For medical care use qualified professionals, not the student model.

Official reference: Understand the difference between a case, infection and exposure. Verify the current status and reporting date before quoting live information.

Build a hypothetical sensitivity-and-specificity table

Where: An indoor worksheet with entirely fictional screening results.

Why this is an epidemiology question

Epidemiology students encounter tests whose sensitivity and specificity differ. Sensitivity concerns the fraction of truly affected people a test identifies, while specificity concerns the fraction of truly unaffected people it correctly identifies. Neither alone tells a person the probability they have a condition after a result; prevalence and testing context matter. A Punggol school project can demonstrate this with a fictitious two-by-two table without collecting medical data or making claims about any named clinical test.

How to do the exercise

Draw a two-by-two table with a fictional reference classification and simulated test result. Label true positive, false negative, false positive and true negative cells. Provide a small whole-number sample, calculate sensitivity and specificity with the correct denominators, then identify which values a clinician or researcher would need from a properly validated study. Keep the disease and test generic so nobody reads the numbers as medical advice.

Worked example

Fictional test data: 80 reference-positive cases contain 72 positive results and 8 negative results; 120 reference-negative cases contain 12 positive results and 108 negative results. Sensitivity is 72/80 = 90%; specificity is 108/120 = 90%. Yet among all 84 positive results, only 72 are reference-positive in this hypothetical sample. The apparent positive predictive fraction is 72/84, about 85.7%, not 90%. The calculation shows why denominators matter.

What to investigate next

Change only the reference-positive prevalence while keeping the hypothetical sensitivity and specificity fixed; discuss how the fraction of positive tests that are true positives can shift. Label assumptions and use an appropriately constructed table so rounding does not produce impossible people counts. Advanced students can explore confidence intervals but should not present their numbers as evidence for a real product.

Ethics and data limits

Do not apply these fictional performance figures to any COVID, dengue or other actual test. Screening, diagnosis and counselling are clinical matters subject to professional standards and official guidelines.

Official reference: Build a hypothetical sensitivity-and-specificity table. Verify the current status and reporting date before quoting live information.

Compare regional PM2.5 readings and personal exposure claims

Where: NEA myENV and data.gov.sg’s major-region PM2.5 API.

Why this is an epidemiology question

NEA air-quality information includes one-hour PM2.5 and twenty-four-hour PSI measurements with different meanings. Data.gov.sg’s documented PM2.5 API returns hourly readings for major reporting regions, not a block-by-block sensor map. An epidemiological exposure model must be careful about geographic and temporal aggregation. Calling an entire region’s hourly value ‘the concentration every Punggol child inhaled’ would produce false individual precision, even if the regional number were authentic.

How to do the exercise

Open the official data.gov.sg specification and note its region, timestamp and unit. Make a small fictional three-hour series and list one hypothetical student’s actual possible environments: home, sheltered station, school classroom. Ask whether the regional reading establishes conditions inside each location. Plot the numbers as a regional time series with a clear source label, never as the student’s personal exposure. Read current NEA advisories before any actual outdoor activity.

Worked example

Fictional three-hour regional values are 15, 20 and 25µg/m³. Their arithmetic mean is 20µg/m³, but it does not represent indoor personal exposure or an individual’s biological response. A chart titled ‘Punggol children breathed 20µg/m³’ would be misleading. A correct title says ‘Illustrative three-hour major-region readings, not personal data.’ The change is methodological, not merely cosmetic.

What to investigate next

Introduce exposure misclassification and ecological fallacy. A regional average might be useful for population-level context but does not describe every individual’s exposure pattern. Ask which data would be needed for a genuine personal study and why consent, sensitive information safeguards, calibrated instruments and qualified oversight would matter.

Ethics and data limits

Do not expose students deliberately to haze or use outdated screenshots as real-time health warnings. The example values are invented; real air-quality decisions belong to NEA’s current advisories.

Official reference: Compare regional PM2.5 readings and personal exposure claims. Verify the current status and reporting date before quoting live information.

Explain selection bias using two Punggol public locations

Where: Punggol Waterway Park and Punggol Regional Library as hypothetical sampling sites.

Why this is an epidemiology question

If a student asks only people in a park about outdoor exercise, the responses may overrepresent people who were willing and able to go there. Asking only library visitors about reading time creates a different selection effect. Punggol’s real public spaces make that problem easy to visualise, but nobody needs to approach actual visitors to learn it. Epidemiology and population research require matching the sampling frame to the target population.

How to do the exercise

Write a fictional research question: ‘What fraction of all Punggol residents undertakes outdoor recreation on a typical weekend?’ Identify the target population, the proposed convenient sample and who might be systematically absent. Invent response cards from different groups without attaching real names. Compare the bias that follows when interviews are taken only on a sunny Saturday at Waterway Park versus at a Monday library event. Propose how a more responsible study might use probability sampling or existing aggregated data.

Worked example

A fictional student interviews ten park visitors, nine of whom say they enjoy park walks, and publishes ‘90% of Punggol residents enjoy outdoor exercise’. The sample includes people already using the park; it cannot support the general population claim. A corrected report says ‘Nine of ten people in this invented convenience sample expressed the preference’ and clearly notes the design’s limitations. A real study would need appropriate consent and representative methods.

What to investigate next

Consider non-response bias and recall bias. Some people may be absent at the chosen time; others might have trouble remembering how often they exercise. Ask which kinds of questions or record-based alternatives could reduce those problems while preserving privacy. Emphasise that bigger convenience samples do not automatically remove systematic bias.

Ethics and data limits

Do not solicit children, patients, elderly residents or other public visitors as unsupervised research subjects. Conduct the initial exercise with wholly fictional cards. Formal human-subject research may require school or institutional review and consent.

Official reference: Explain selection bias using two Punggol public locations. Verify the current status and reporting date before quoting live information.

Distinguish an association from a cause in a housing-and-health graph

Where: HDB public Eco-Town information and aggregate planning-area research.

Why this is an epidemiology question

Punggol’s Eco-Town includes green design, public walking routes and housing planned around urban amenities. It may be tempting to claim that a shaded path causes better population health simply because places with greenery appear attractive. A correlation between green coverage and activity in a hypothetical dataset does not prove a specific individual or population health benefit. Age, work, disability, weather, nearby services and other factors can influence both exposure and outcome.

How to do the exercise

Use HDB’s published green-feature descriptions to establish what was designed. Then draw a causal diagram for an entirely fictional study comparing neighbourhood greenness and outdoor walking frequency. Put possible confounders on a separate row—household composition, accessible transport, work schedules, heat and income—and draw arrows showing how each might affect the observed association. Record what the HDB page verifies and what remains a causal hypothesis requiring proper study.

Worked example

A class compares two hypothetical zones: one with more trees and higher reported walking, another with less tree cover and lower walking. The student writes, ‘Adding trees definitely caused the difference.’ The corrected analysis says the association could be consistent with a beneficial environment, but without suitable controls or study design it is not proof of cause. The next step is identifying confounders and appropriate evidence, not adding more decimal places.

What to investigate next

Discuss how natural experiments, before-and-after data and comparison groups can improve causal inference while still requiring caution. Identify which public environmental interventions are chosen by planners and which resident characteristics are not randomly assigned. The student should learn to state a plausible question rather than perform unauthorised research.

Ethics and data limits

Do not collect residents’ medical or income records, diagnose people from location, or treat HDB’s published design goals as independently measured health impacts. Stay in lawful public areas when observing infrastructure.

Official reference: Distinguish an association from a cause in a housing-and-health graph. Verify the current status and reporting date before quoting live information.

Prepare a date-aware outbreak information brief with neutral language

Where: Home or classroom using current NEA and MOH primary information.

Why this is an epidemiology question

Epidemiology findings are easily distorted when an old case count is forwarded as current or a neighbourhood is singled out without evidence. NEA’s live cluster map is updated and distinguishes red, yellow and green statuses. A good public-health brief should state location scope, date, evidence and practical guidance without amplifying a rumour. Ethical communication avoids identifying infected residents or inventing risk for a particular Punggol block.

How to do the exercise

Choose a fabricated scenario clearly labelled exercise-only, with invented cases and an abstract map. Draft a short brief that begins with verified facts from NEA about how its reporting system works, explains which hypothetical numbers were assumed and directs readers to the current official map for actual status. Add ‘last reviewed’ and a plain-language definition of the alert level. Ask a peer to identify whether any sentence might be mistaken as a real warning; revise those words before sharing in a class setting.

Worked example

The first worksheet title says ‘Punggol is a red dengue zone today!’ although all numbers are hypothetical. The corrected title says ‘Practice exercise: how NEA classifies an illustrative cluster’ and removes any real street names. Inside the brief, the student uses actual NEA definition and links to current updates. This preserves the useful epidemiology lesson without alarming residents.

What to investigate next

Create a correction log for a hypothetical outdated chart: original timestamp, incorrect inference, corrected definition and source date. Ask why the right response to a genuine urgent warning is to check authoritative updated information rather than debate the original post’s popularity. Compare a calm accurate headline with a fear-based one and discuss public trust.

Ethics and data limits

Never produce realistic fabricated disease alerts for publication, impersonate MOH/NEA, or disclose sensitive case details. Genuine illness concerns should go through healthcare and public-health channels rather than a student’s unpublished analysis.

Official reference: Prepare a date-aware outbreak information brief with neutral language. Verify the current status and reporting date before quoting live information.

Write a mini epidemiological study proposal without collecting patient data

Where: Private study space, with Punggol as a documented community context.

Why this is an epidemiology question

A study proposal integrates the skills of defining a population, framing an outcome, choosing a design, planning analysis and considering ethics before research. Punggol has public population and environmental information suitable for desk exercises, but student curiosity is not permission to seek medical records or conduct clinical surveys. A carefully designed proposal can be a strong educational result without observing any actual patient or accessing a clinic.

How to do the exercise

Choose a narrow non-clinical topic such as ‘How might shared-path access relate to self-reported walking routines?’ Draft a hypothetical objective, target population, variables, sampling frame, measures, confounders and a proposed descriptive analysis. Mark all human participation as requiring informed consent and appropriate institution review. Use only published aggregate sources and fictional values in the worked example. Clearly state what the proposal cannot infer from a cross-sectional design and what permissions a real study would need.

Worked example

An initial proposal says, ‘Ask anyone at Oasis Terraces Polyclinic whether they have dengue and photograph their address for mapping.’ It would intrude on sensitive health information and is methodologically biased. The revised beginner proposal uses a published NEA surveillance description, SingStat population data and simulated anonymised records. No patients are contacted, and the student’s main result is a transparent design and ethics checklist.

What to investigate next

Turn the proposal into a one-page preregistration-style note recording the primary outcome, sample-size rationale to be developed professionally, analysis and limitations. Discuss why changing a hypothesis after viewing the data can exaggerate apparent findings. A research design is strengthened when it can be reviewed before any data collection.

Ethics and data limits

No clinic or care facility permits unrestricted student epidemiology research. Confidential medical data, genomic information and patient identities require strict legal, institutional and professional safeguards. This is not a medical study or a diagnosis.

Official reference: Write a mini epidemiological study proposal without collecting patient data. Verify the current status and reporting date before quoting live information.

Three manageable epidemiology study sessions

A case-definition workshop

Use NEA’s dengue cluster criteria and three fictional case cards, writing down onset dates and distances. Ask which pairs meet the stated operational timing-and-distance thresholds. The answer should refer to those specific assumptions and acknowledge that real NEA classification requires authoritative case records and assessment. Finish by rewriting a sensational unverified headline into a factual training label.

A population-data desk afternoon

Read the current SingStat geographical distribution source and NEA’s case-reporting explanation. Do not combine figures merely because both mention Singapore. Choose a completely fictitious district population and number of cases, calculate a rate per ten thousand, and identify what would need to be true to compare it with another group. Test one error—wrong denominator, mismatched month or duplicated report—and repair the calculation.

An environmental exposure and research-ethics session

Open data.gov.sg’s PM2.5 API documentation and distinguish major-region readings from a person’s actual indoor exposure. Make a fictional time series with timestamps and units. Ask whether a park-versus-library convenience sample would represent every Punggol resident. End with a one-page study proposal using only public or simulated data, explicitly listing the ethics review and consent needs for any future real research.

A field-note and data-quality checklist

• Specify an exact event or outcome definition.
• Record whether dates refer to symptom onset, notification or publication.
• Define the relevant geography and time period.
• Identify an appropriate population denominator.
• Distinguish count, rate, risk, prevalence and cumulative total.
• Check for missing, revised or duplicate records.
• Label invented practice values unmistakably.
• Consider selection bias, reporting bias and confounding.
• Avoid individualised health predictions from regional data.
• Protect confidentiality and obtain institutional permission before any human research.

Frequently asked questions

What can Epidemiology students study in Punggol without patient data?

NEA surveillance criteria, SingStat population denominators, major-region air-quality data, observational biases and hypothetical study designs.

What does NEA mean by a dengue cluster?

Its operational criteria include at least two cases with symptom onsets within fourteen days and relevant addresses within 150 metres.

Does a cluster map mean every home inside it has dengue?

No. It identifies a locality for public-health intervention, not the illness status of every resident.

What is NEA’s red dengue alert category?

The current map labels high-risk clusters with ten or more cases as red; yellow refers to fewer than ten, and green indicates a surveillance status.

Can the cluster list change daily?

Yes. Use the official current map and its listed update time rather than an old screenshot.

What is incidence?

New cases or events occurring in a defined population during a specified time, often presented as a count or rate.

What is prevalence?

The proportion or count of existing cases in a defined population at a specified time or period under a stated definition.

Why do weekly dengue counts differ from cumulative totals?

Weekly counts measure new reported cases in a period; cumulative totals add reports across periods and can rise while weekly counts fall.

Why do epidemiologists use denominators?

To compare events relative to population size and a compatible period instead of ranking places only by raw case counts.

Does SingStat have 2026 Punggol population data?

Yes. Its published planning-area and subzone data provide demographic denominators, with geographic and reference-date definitions.

Can an exposure be equated with confirmed illness?

No. Possible exposure, infection, symptoms and case reporting are distinct states.

Can a student diagnose dengue from a mosquito sighting?

No. Symptoms and diagnosis require appropriate healthcare assessment.

What is sensitivity?

The fraction of reference-positive individuals correctly identified by a test in a defined validation context.

What is specificity?

The fraction of reference-negative individuals correctly identified by a test in a defined validation context.

Do 90% sensitivity and 90% specificity imply a 90% chance of illness after a positive result?

No. Predictive values also depend on prevalence and other contextual factors.

Does regional PM2.5 measure the air inside a Punggol flat?

No. A region-level reading cannot be taken as a personal indoor exposure measurement.

What is selection bias?

A systematic difference between the selected sample and the intended population that can distort inference.

What is confounding?

A factor associated with both the exposure and outcome that may distort an observed association.

Can I interview polyclinic patients for a class project?

Not without appropriate institutional permission, informed consent and health-data safeguards. Use simulated or published aggregate data for beginner exercises.

Can I publish a practice outbreak map?

Only if it is unmistakably labelled fictional and cannot be mistaken for a real local health warning.

Is this guide medical advice?

No. It is an educational guide to research reasoning and public-data interpretation; real health concerns need qualified care.

Official sources and related Punggol guides

• Decode NEA’s actual dengue cluster definition
• Separate new dengue cases from cumulative cases
• Learn the incidence-rate denominator from SingStat’s population tables
• Understand the difference between a case, infection and exposure
• Build a hypothetical sensitivity-and-specificity table
• Compare regional PM2.5 readings and personal exposure claims
• Explain selection bias using two Punggol public locations
• Distinguish an association from a cause in a housing-and-health graph
• Prepare a date-aware outbreak information brief with neutral language
• Write a mini epidemiological study proposal without collecting patient data
• NEA dengue clusters under surveillance
• SingStat Population Trends 2026
• Related Punggol Public Health learning
• Related diagnostic science and measurement ethics
• Related statistics and sampling examples
• eduKateSG worked-example learning and correction standard

Reliable public-health reasoning starts with a correct denominator

Epidemiology is not about diagnosing strangers from a map or making a sensational local headline. It is about clearly defining events and populations, choosing a defensible method, recognising uncertainty and telling readers exactly what the evidence supports. Punggol offers real official surveillance, demographic and environmental data for studying these ideas, while the safest first projects can remain entirely on paper.