Degdeg

Frameworks & Standards

Degdeg's design choices are informed by, and designed in alignment with, a set of international frameworks for early warning, climate-health decision support, and disaster risk reduction. This page states plainly what that alignment means in product terms — and, just as importantly, what it doesn't: Degdeg is not certified, endorsed, accredited, or in partnership with any of the bodies below, and does not implement any of their scoring methodologies.

For deeper technical detail on how the AI pipeline itself is grounded, validated, and audited, see the Methodology page.

Early Warnings for All (EW4All)

UN initiative, co-led by WMO and UNDRR, with ITU and IFRC

A UN-wide push for every person on Earth to be covered by a life-saving early warning system by the end of 2027 — today, only around 40% of Africa has that coverage.

How Degdeg Reflects This

  • We consume, not produce, satellite Earth-observation data and forecasts — Degdeg sits downstream of detection and forecasting, not inside it. This includes Open-Meteo (rainfall forecasts), CHIRPS (observed rainfall), NDVI (vegetation condition), GloFAS (forecast river discharge), and ICPAC's Combined Drought Indicator — ICPAC (the IGAD Climate Prediction and Applications Centre) is a WMO-accredited regional climate centre. We use their openly-licensed published data — this is consumption, not a partnership.
  • We now contribute more substantially to Pillar 1 (risk knowledge): Earth-observation data from CHIRPS, ICPAC, NDVI, and GloFAS is processed into woreda-level indicators, combined with forecast data, to produce anticipated health risk grounded in a cited rules table.
  • We contribute to Pillar 3 (warning dissemination and communication) with advisories drafted in Somali and English across WhatsApp, radio, and formal-memo formats.
  • We contribute to Pillar 4 (preparedness and response) through the Recommended Actions layer and the alert lifecycle (created, acknowledged, action taken, outcome recorded).
Official Source: earlywarningsforall.org
WHO Operational Framework for Building Climate Resilient and Low Carbon Health Systems

World Health Organization, 2023

Ten components for climate-resilient health systems, mapped to the six WHO health-system building blocks.

How Degdeg Reflects This

  • Integrated risk monitoring and early warning systems — Degdeg's core function.
  • Climate-informed health programmes — the impact rules table links hazard conditions to specific disease/nutrition programme responses.
  • Health information systems — an append-only hazard/health-risk/outcome data model and a full audit log of every AI call.

What's Out of Scope

  • We do not address low-carbon health facilities, sustainable financing, or health workforce climate training — those components of the framework are outside Degdeg's current scope.
Official Source: who.int
WMO Guidelines on Multi-hazard Impact-based Forecast and Warning Services

World Meteorological Organization

The shift from forecasting "what the weather will be" to "what the weather will do" — incorporating exposure and vulnerability, not just hazard magnitude.

How Degdeg Reflects This

  • This is our core thesis: Stage 2 (anticipate) of the pipeline exists specifically to translate a hazard level into an anticipated health impact, not just report rainfall or drought severity.
  • Every anticipated impact cites a specific rule linking hazard condition, health outcome, lead time, and pathway — not a generic severity label.
  • We now combine observed Earth-observation evidence (CHIRPS, NDVI, ICPAC) with forecast data (Open-Meteo, GloFAS) — closer to this guidance than an assessment built on forecast alone.

What's Out of Scope

  • Honestly: we do impact-based translation, not yet vulnerability-weighted risk. Our vulnerability/exposure layer (population, health-system capacity, WASH, baseline disease burden) doesn't exist yet — see "What We Don't Claim Yet" below.
Official Source: library.wmo.int
WHO Climate and Health Training Modules — Module 5 (Assessment and Prediction of Health Impacts)

World Health Organization

WHO's own reference material on decision-support over prediction, alert thresholds, weekly bulletins, prospective validation, and linking forecasts to response actions — the playbook Degdeg's design already draws on.

How Degdeg Reflects This

  • Live today: hazard/risk levels (low/watch/warning/emergency), the Recommended Actions layer, and the full alert lifecycle (created → acknowledged → action taken → outcome recorded).
  • Live today: forecast history is preserved permanently and never overwritten — a prerequisite for any future validation.
  • Architected but awaiting data: prospective validation / forecast performance metrics. The outcome-storage container exists (`health_outcomes`), but we publish no accuracy figure until real surveillance data supports one — see docs/verification-readiness.md.
Official Source: who.int
INFORM Climate Change

European Commission Joint Research Centre (JRC)

A conceptual framework and composite risk index built from hazard, exposure, vulnerability, and coping capacity.

How Degdeg Reflects This

  • We do not implement INFORM or compute an INFORM score of any kind.
  • It informs how we think about the difference between a hazard and a risk: our hazard_states → health_risks data model reflects exactly that distinction, kept as two separate, explicitly linked tables rather than one collapsed "risk level."

What's Out of Scope

  • The exposure/vulnerability/coping-capacity side of INFORM's model is the part we haven't built — see "What We Don't Claim Yet" below.
Official Source: drmkc.jrc.ec.europa.eu
Sendai Framework for Disaster Risk Reduction 2015–2030

UN Office for Disaster Risk Reduction (UNDRR)

The umbrella global agreement on disaster risk reduction that EW4All sits under, and that impact-based forecasting serves — understanding risk before it becomes a disaster.

How Degdeg Reflects This

  • Referenced here as context, not a framework Degdeg implements directly — EW4All (above) is the operational initiative pursuing Sendai's early-warning priority, and that's the level at which Degdeg actually participates.
Official Source: undrr.org
Ethiopia's National Artificial Intelligence in Healthcare Programs Technical Working Group

Ethiopia's national AI-in-health governance process

A national technical working group coordinating artificial intelligence programs and policy for Ethiopia's health sector.

How Degdeg Reflects This

  • AHRI AI Innovation Lab leadership sits on this Technical Working Group, and is also an expert member of CHOICE (the Coalition for Health AI Innovation and Ethics, an Africa CDC initiative led by AHRI) — direct visibility into national AI-in-health governance, not just external framework alignment.
  • This is an institutional fact about who develops Degdeg, not a claim that Degdeg itself has been reviewed or endorsed by the Working Group or CHOICE.
Official Source: ai.ahri.gov.et

What We Don't Claim Yet

Transparency about what's still in development is a credibility signal, not a weakness — here is what Degdeg does not yet do:

  • Vulnerability/exposure weighting (population, health-system capacity, WASH, baseline disease burden, food security, displacement) — anticipated risk is currently hazard-driven, not yet adjusted for who is exposed and how able they are to cope.
  • Forecast performance metrics (accuracy, calibration, lead-time validation) — the outcome-storage container exists, but is empty by design until real surveillance data accumulates — see docs/verification-readiness.md.
  • Multi-channel alert delivery (SMS, email, automated WhatsApp dispatch) — alerts are visible to signed-in staff on the dashboard today; no delivery integration exists yet.
  • DHIS2 integration — no connection to Ethiopia's national health information system exists yet; this would be the natural path for surveillance data to enter automatically instead of by hand.
  • Soil moisture and land surface temperature — identified candidate Earth-observation sources, not yet integrated into the data used.
  • GloFAS return-period thresholds — not yet computed; flood assessment confidence never rises above low as a result.
  • Observed flood extent (e.g. Copernicus Global Flood Monitoring) — evaluated and deliberately deferred, not ingested.
See the Methodology page for the technical detail