Build a Global Health Indicator Dashboard

A global health dashboard becomes unreliable when it treats every indicator as a simple country-year number. WHO observations can include sex and other dimensions, estimates, bounds, and different update schedules. The WHO Global Health Observatory Data Scraper preserves those attributes for a governed reporting layer.

Build an indicator registry first

Expose methodology near the visualization instead of burying it in a notebook. Users should see the indicator definition, selected dimensions, newest available year, uncertainty, and source. Those details reduce support questions and discourage comparisons that the observations cannot sustain.

Create a table containing indicator code, official title, unit, preferred dimension filters, interpretation note, and dashboard label. Then use bounded inputs for the countries and years the product actually needs.

{"indicatorCodes":["WHOSIS_000001","WHOSIS_000002"],"countryCodes":["USA","IND","BRA","ZAF","JPN"],"yearFrom":2015,"yearTo":2026,"maxResultsPerIndicator":500}

In the warehouse, key rows on indicator, country, year, and all dimension codes. Keep numeric value, formatted value, lower and upper bounds, WHO update date, and source URL. A latest-value view can power the interface, while the raw table remains available for historical and revision analysis.

Dashboard quality checks should detect duplicate keys, unexpected dimensions, missing country names, inverted bounds, and new observations older than the displayed year. Show data freshness per indicator rather than one global “updated” timestamp. Avoid filling gaps with zero. If interpolation or modeled comparisons are added, label them as derived fields and keep them separate from WHO-published values. This architecture allows the dashboard to expand without sacrificing provenance.