Health Equity Radar

From local health data to local action.

Health Equity Radar helps local public-health teams see which health indicators in their community deserve closer review, why, and which official guidance they can consult. It uses official Korean data and transparent rules.

It is a decision-support and screening tool for identifying indicators that deserve closer review. It does not diagnose, predict or prescribe.

Current stage: Working prototype · detailed deployment in the Republic of Korea · data verified 2026-10-26

90-second tour

Real screenshots of the running site, with a link to open each screen. Korea is the detailed deployment; the last step shows the same framework applied to open global data.

  1. 1 · FIND A COMMUNITY

    Choose a municipality or health-centre area

    Search any of the 229 municipalities or 258 survey units. The example is Gangneung-si (Korean home screen).

    Open the home screen →
    Korean home-screen card ‘Quick look at local health gaps’ with ‘Gangneung’ typed in the search box and three mini cards: an indicator to review first, a recently improved indicator and an indicator with large regional gaps.
  2. 2 · IDENTIFY WHAT TO REVIEW FIRST

    Up to three indicators, one per domain

    Each indicator shows the local value, the national median, the relative position, the recent trend and deprivation context. Tiers: Review first, Monitor, Relatively favourable, Insufficient data.

    Open this screen →
    Health Equity Priority card for Gangneung-si in English: counts by tier and the first indicators flagged for review with local value, national median, relative position, recent trend and deprivation context.
  3. 3 · SEE WHY

    Reasons computed from the data

    Gap from the national median, position among comparable areas, recent trend, deprivation (side by side, never as a cause) and whether the 95% confidence interval excludes the median.

    Open this screen →
    ‘Why it matters’ list for the first flagged indicator: difference from the national median, position among survey units, five-year trend, deprivation quintile and confidence-interval check.
  4. 4 · REVIEW POSSIBLE ACTION

    Official guidance, not generated advice

    Links to Korean national and provincial plans, WHO, NICE and CPSTF documents, with collection dates. If nothing is linked, the card says so.

    Open this screen →
    ‘Possible public-health action’ list for the first flagged indicator: linked provincial and national plan items, NICE guideline and CPSTF recommendations, each with a link to the original document.
  5. 5 · SCALE THE FRAMEWORK

    Same rules, open global data

    The global prototype applies the framework to World Bank data from 217 countries and economies. Current position and observed historical trend are shown separately.

    Open the global prototype →
    Global prototype signal card for the Republic of Korea: current position compared with high-income economies on one side and the observed historical trend since 2000 on the other, with a note that trends do not explain causes or predict future values.

What is real today

Counted from the data files the site loads (Competition Evidence Pack, 2026-10-26). Numbers at different geographic levels are not added or multiplied.

Korea detailed local deployment

171health indicators9 source groups
41survey indicatorsKorea Community Health Survey, across 258 survey units
130administrative / derived indicatorsacross 229 municipalities
2008–2025years17 provinces; coverage differs by source

Global prototype portability test

217countries and economiesprototype applied to their published data
25World Bank WDI indicatorsstandardized open data (CC BY 4.0)
115,457observed values2000–2025; observed years only, no interpolation
21evidence links checkedaddress and title verified; expert relevance review pending

Automated checks: All passed — 26 unit tests; 7 browser test suites with 411 checks (last run 2026-10-26 17:00 KST, commit 78c6f11).

What it does not do

It is a decision-support and screening tool for identifying indicators that deserve closer review.

  • Does not diagnose disease
  • Does not predict future disease
  • Does not estimate causal policy effects
  • Does not automatically recommend public-budget allocation
  • Does not claim neighbouring ranks are meaningfully different
  • Does not replace local public-health judgement

Built in Korea. Tested as a portable framework globally.

Korea — detailed local deployment

  • 41 survey indicators across 258 survey units and 130 administrative / derived indicators across 229 municipalities, 2008–2025.
  • Priority, reasons and linked official guidance for one community at a time.
  • Survey uncertainty (95% confidence intervals, relative standard error) built into the rules.

Global prototype — portability test

  • Prototype applied to data from 217 countries and economies, using 25 standardized WDI indicators.
  • Each value is compared with economies in the same World Bank income group.
  • Observed historical trends shown separately from the current position; no interpolation.
  • Evidence links technically verified (21 sources); expert relevance review still pending.
  • National averages only: it does not show differences within countries.

The global prototype is not deployed in, used by or endorsed by any country or organisation. Global methodology

The problem

In Korea, the data a local public-health team needs for one area are published by different agencies in different systems: the Korea Community Health Survey (Korea Disease Control and Prevention Agency), cause-of-death statistics (Statistics Korea), and screening and insurance statistics (National Health Insurance Service). Each has its own tables, years and geographic units — the survey is reported for 258 survey units, most administrative statistics for 229 municipalities.

Local governments prepare four-year local health plans; preparation of the next cycle (2027–2030) begins in the second half of 2026. Comparing an area across sources means opening several systems, reconciling units and years, and judging whether a difference is larger than survey uncertainty.

How priorities are identified

For each indicator the tool combines four components scored 0–1 and averages those available. Missing parts are dropped and the weights renormalized; they are never set to zero.

  • Gap (35%) — direction-aware relative difference from the national median.
  • Relative position (25%) — share of comparable areas with a better value.
  • Recent trend (20%) — five-year slope, by tertile of the comparison group.
  • Socioeconomic context (20%) — area deprivation quintile, shown side by side, never as a cause.

“Review first” also requires the lower 25%, a score of at least 0.55, a 95% confidence interval that excludes the national median, and a relative standard error of 20% or less. Only tiers are shown, not scores. No prediction models or causal inference are used. Full rules: English overview · data validation · indicator metadata.

Data sources

SourceRole in the toolIndicators
Korea Disease Control and Prevention Agency (KDCA) — Korea Community Health Survey, via KOSISHealth behaviours, chronic-disease management, prevention, mental and oral health; survey-weighted rates with standard errors (used for confidence intervals)41
Statistics Korea — Cause-of-death statisticsAge-standardized mortality by cause22
National Health Insurance Service (NHIS) — National cancer screening statisticsCancer screening rates (people screened ÷ people eligible)7
National Health Insurance Service (NHIS) — health insurance, health screening and medical-use statisticsScreening, medical use and health-workforce context33
KDCA — Notifiable infectious disease reportsInfectious disease incidence12
Statistics Korea, Ministry of the Interior and Safety and othersPopulation, social and economic context24
Ministry of Environment, Ministry of Land, Infrastructure and Transport, Korea Road Traffic Authority and othersEnvironment and safety context28
Derived in this project from Statistics Korea life tables and the Community Health SurveyHealthy life expectancy (approximation, not an official statistic)3
Derived in this project from the Population and Housing Census (2015, 2020)Area deprivation index (approximation) — socioeconomic context1

Most administrative indicators were obtained from their original producers via a public research database (Kim D. et al., v1.7, KDCA archive); mortality 2018–2025 was re-checked against Statistics Korea tables. Linked guidance: Korean national and provincial health plans, WHO, NICE (UK) and the US Community Preventive Services Task Force.

Evidence of use and next validation

No user-testing or impact evidence yet, and no usage figures are claimed. External review requests have been sent; no reviews have been published yet.

  • Planned: a structured test with 5–8 local public-health practitioners (find their area, identify one priority, explain why, open the evidence, compare with another area).
  • The in-app “Give Feedback” form records answers locally (copy, email or .json/.txt file); nothing is sent to a server.
  • Independent methodological review, published only with the reviewer’s consent.

Known limitations

  • Korean interface except this page, the English overview and the Health Equity Priority card.
  • Survey indicators carry sampling error; neighbouring ranks may not differ meaningfully.
  • No programme input or output data, so the tool cannot evaluate programme performance or effects.
  • Healthy life expectancy and the deprivation index are approximations computed for this project, not official statistics.
  • The prevention knowledge base was collected and summarized with AI assistance; original documents are linked and should be checked.