Transmission risk levels - methodology

1. Purpose of this article

This article explains what GIDEON’s transmission risk levels 1-4 mean, what informs them, and how baseline and weather-informed levels relate. It is written for customers, institutional reviewers, and partners who need to interpret or cite GIDEON maps in clinical, travel-medicine, or public-health workflows.

It describes GIDEON’s methodology through level definitions, evidence sources, curation policy, quality review, and limitations. It intentionally does not disclose proprietary scoring weights, rule thresholds, numeric weather calibration, or full algorithmic specifications.

Scope: Chikungunya, Dengue, Malaria, Yellow fever, and Zika transmission risk maps (may expand to additional diseases).

2. Summary

GIDEON publishes transmission risk on a four-level ordinal scale (1 = lowest, 4 = highest) for selected mosquito-borne diseases.

For each location and disease, the product may show:

Baseline answers: Can this disease transmit here under supportive conditions?
Current answers: Are recent environmental conditions likely suppressing transmission relative to that baseline right now?

The same ordinal scale and general methodology apply across all diseases in scope; which external sources are consulted varies by disease (see  Section 5.2). 

3. The four risk levels

Levels are ordered categories, not precise probabilities. Level 3 is “more risk” than level 2, but it is not calibrated as “75% of level 4.” Cross-disease comparison should be interpreted cautiously.

User-facing map message (e.g. “High risk - …”) is generated from the assigned level and the dominant epidemiological context for that location. Messages avoid displaying raw counts or rates even when those informed the assignment. 

4. Diseases and geographic coverage

4.1 Diseases covered

Transmission risk maps in this family currently include (may expand into additional diseases):

• Chikungunya
• Dengue
• Malaria
• Yellow fever
• Zika

4.2 Geographic units

Risk is assigned to countries and, where data and curation support it, states/provinces, districts, and smaller administrative areas. Country-level ratings apply where subnational detail is not available.

Elevation (where relevant, especially for malaria) may further constrain how risk is described for high-altitude zones.

4.3 Seasonal (monthly) view

Where available, GIDEON may also show month-by-month risk levels based on long-term average climate for each month. This reflects typical seasonal patterns, not day-to-day weather. It is separate from the current weather adjustment described in Section 6.

Levels for unfavorable months may be lower than the annual baseline; the annual baseline itself is not increased by this pass. Specific adjustment rules are proprietary.

5. How baseline risk is determined

Baseline risk is produced by a repeatable, multi-step curation and analytics process.

At a high level:

Authoritative country (and sometimes subnational) reference levels (CDC / WHO / national health ministries)

↓

GIDEON surveillance & epidemiology enrichment

↓

Structured rule-based assignment → baseline level

↓

Geographic consistency checks & expert overrides

↓

(Optional) adjustment for recent weather → current level

↓

Publication to the live product

5.1 External reference inputs

Country-level (and sometimes subnational) ratings from CDC and WHO travel-disease resources (and comparable national guidance) are incorporated as reference context:

These references anchor national/subnational context and quality review. Subnational baseline levels are determined primarily from GIDEON’s own evidence synthesis, not by applying the country rating uniformly to every district.

For some diseases, countries not listed in a given external source may receive a default country reference classification according to GIDEON’s import policy for that disease.

5.2 Evidence from GIDEON

Before levels are assigned, each location is enriched with signals from GIDEON’s infectious-disease knowledge base, including:

From these inputs, GIDEON derives indicators such as whether an outbreak is ongoing, whether activity occurred within the past five years, relative burden per population or per area, survey positivity, and whether elevation makes transmission unlikely.

5.3 Structured assignment

Baseline levels are assigned by a versioned, priority-ordered library of epidemiological rules maintained by GIDEON’s science team. The library contains many rules (on the order of hundreds), each assigning a target level (1-4) when its conditions are met.

Rules are designed to cover the major epidemiological situations GIDEON encounters, including for example:

• Elimination or no local transmission: areas where disease is absent, imported only, or officially classified as non-endemic
• Sporadic or low-level activity: historical or focal transmission without sustained burden
• Endemic presence with moderate signals: ongoing transmission context without the highest-tier triggers
• High burden or active outbreaks: ongoing outbreaks, elevated incidence or survey burden, or dense subnational activity
• Geographic and ecological context: admin level (country vs province vs district), elevation suitability, and related terrain factors

Each rule links a target level to combinations of the indicators in Section 5.2. Rules are evaluated in priority order; the first applicable rule sets the baseline level for that location.

Rules are updated as science and surveillance evolve. Individual rule definitions, numeric thresholds, and exact priority ordering are proprietary.

Default when no rule matches

In practice, almost every location matches at least one rule given the size and scope of the library. In the unusual case that no rule applies, GIDEON assigns a default of level 2 (Low risk).

This default is a neutral, middle posture for data-sparse edge cases:

• It does not assign Unlikely (1) — GIDEON does not treat “no rule matched” as proof of absence of transmission.
• It does not assign High (4) — without evidence, the map does not imply active high burden.
• It is not the same as copying the country-level CDC/WHO reference to every subnational unit; country references inform context (Section 5.1) but are not automatically applied as the district-level score when rules do not fire.

Why level 2: Level 2 is used so that a rare “no rule matched” outcome is not read as Unlikely (1) (absolute absence of transmission) or inflated to High (3-4) without epidemiological triggers in the rule library. It reflects insufficient rule match, not a confirmed low-burden assessment for that location.

Separate national consistency logic (below) can still adjust subnational levels upward when a lower local assignment would contradict the country-level assessment.

National consistency: When a subnational area would be classified as “unlikely” but the country-level assessment is higher, the subnational level may be adjusted upward to stay consistent with national context.

5.4 Evidence hierarchy and resolving conflicts

When multiple sources disagree, GIDEON applies the following general principles (not every exception is listed).

Source precedence (typical)

1. Recent, location-specific official surveillance (national ministry bulletins, subnational tables, WHO/PAHO assessments with explicit geography)
2. GIDEON-curated outbreaks and case trends at matching administrative granularity
3. Seroprevalence surveys meeting GIDEON criteria for active transmission (human populations, appropriate test types, recent field dates)
4. Country-level CDC/WHO travel-disease classifications (anchor national context; not automatically copied to every district)
5. Ecological exclusion (e.g. dominant land area above vector altitudinal limits where relevant)

Temporal precedence

• Ongoing outbreak activity generally weighs more heavily than historical-only signals for upward classification.
• Recency windows prevent stale events from indefinitely elevating risk; exact windows are proprietary.
• Baseline curation intentionally de-emphasizes isolated news spikes unless corroborated in GIDEON’s curated outbreak record.

Geographic granularity

• Finer-resolution evidence wins over country generalizations when quality and geography are clear.
• If only national guidance exists, subnational units may remain unscored, inherit country context through consistency rules, or receive levels only after granular country programs (Section 5.6).

When GIDEON data and external maps disagree

Documented override source (e.g. ministry report, WHO publication) is retained for audit. Overrides that contradict strong recent GIDEON activity require epidemiologist sign-off.

5.5 Expert review, overrides, and geographic consistency

After automated assignment, GIDEON applies additional quality steps:

Overrides are commonly used where official maps exist but GIDEON surveillance is sparse. They are not applied when they contradict strong recent GIDEON activity without epidemiologist review.

These steps reflect expert judgment; not every override is listed in this document.

5.6 Granular country programs

For selected countries, GIDEON runs dedicated subnational baseline projects in addition to the global pipeline:

Policy: Subnational scores require geography-specific evidence. Country-only generalizations are not automatically applied to every district. Unsupported units remain unassigned or inherit through consistency rules-not silently filled from national burden alone.

Country-specific mapping approaches used in individual programs are not generalized globally and are not disclosed in this document.

5.7 What baseline represents

Baseline is intended as an upper-bound / peak transmission potential for the location under favorable vector and climatic conditions, integrating:

• Authoritative public-health guidance (CDC, WHO, national ministries)
• Historical and recent outbreak activity recorded in GIDEON
• Reported incidence and survey signals
• Endemicity and ecological constraints (elevation, typical seasonality)

A location at High (4) does not assert that transmission is high today; it asserts that transmission can reach that tier when conditions support it.

6. Weather-informed current risk (not yet generally available)

Status: The weather-informed current level is in limited release and is not yet generally available (GA) in all deployments. The following describes the feature when it is enabled in your environment.

For the five diseases listed above, GIDEON may compute a current level from roughly the past 30 days of local weather, using the same global weather summaries that support GIDEON’s environmental features.

6.1 Principles

• Uses daily maximum temperature, relative humidity, and rainfall (including both sustained wetness over the window and a modest adjustment for very heavy recent rain).
• More recent days count more than older days in the window.
• Weather is combined into an overall suitability score, then converted to a factor that can only reduce the baseline level, never increase it.
• The level shown may require consistent unfavorable weather over consecutive update cycles before a decrease is published, to avoid reacting to a single unusual day; return toward baseline when weather improves is not delayed.
• If weather data are missing, current equals baseline.

Malaria uses a temperature suitability profile tuned to malaria vectors; the other weather-adjusted diseases share a profile suited to typical arboviral transmission patterns.

How the three pillars combine (qualitative)

GIDEON does not rely on any single weather variable in isolation. The three pillars are merged into one overall suitability picture for the lookback window:

• Temperature generally carries the strongest influence on the combined picture, because vector activity and pathogen development are strongly temperature-dependent.
• Rainfall is the next most influential factor, reflecting both sustained wetness over the window and very recent heavy rain; the rain signal is combined so that one extreme storm cannot dominate the entire month without support from the broader wetness pattern.
• Humidity contributes additionally, capturing dry or excessively muggy conditions that can suppress typical vector-borne transmission.

If one pillar is missing for a location, the others still contribute; if weather is unavailable altogether, current equals baseline (Section 6.1). Exact blend proportions, suitability curves, and conversion to the displayed level are proprietary.

Absolute suitability, not climate anomaly

For the weather-informed current level, GIDEON evaluates recent observed weather at each location against biological suitability ranges for vector-borne transmission (temperature, humidity, and rainfall). It does not score whether the past month is warmer or wetter than that location’s long-term historical average.

Suitability is assessed against transmission-relevant absolute conditions (for example, temperatures that suppress or support vector activity), not against a percentile of past decades for that district. The separate monthly seasonal view (Section 4.3) uses long-term average climate by calendar month and is independent of this rolling current-weather adjustment.

6.2 Biological rationale

Mosquito-borne transmission depends on conditions that unfold over days to weeks: vector development, breeding-site response to rain, and pathogen extrinsic incubation in the vector. A single day’s weather is therefore a poor proxy for near-term transmission potential. GIDEON uses roughly one month of recent weather as a practical compromise between biological relevance and operational stability — it is not a full entomological simulation. This window also reflects the delayed biological effects of weather on vector breeding and pathogen incubation.

6.3 What you may see in the product

When current is below baseline, GIDEON may show a short note such as:
“Past 30 days, [temperature / rainfall / humidity / weather] patterns lowered transmission risk from [baseline label] to [current label].”

Numerical calibration details (weights, curves, multipliers, persistence settings) are proprietary and may change between product releases.

7. Governance, update cadence, and publication

7.1 Roles

7.2 Production gates

1. Candidate build: full pipeline produces a draft map.
2. Change review: material differences vs prior production are highlighted.
3. Human approval: epidemiology signs off on significant shifts, new overrides, and country programs.
4. Publication: approved map replaces the live product.
5. Weather updates: when enabled, current levels refresh on a shorter cycle without rebuilding baseline.

7.3 Update cadence (typical)

Contact your GIDEON account team for baseline refresh frequency and whether weather-informed current levels are available in your subscription.

7.4 Expedited and ad hoc updates

Scheduled updates are the default for baseline maps. Ad hoc baseline changes may occur when epidemiology publishes curated overrides, new national surveillance publications are imported, or a full pipeline run is executed ahead of schedule after internal review.

GIDEON does not automatically republish baseline levels solely because WHO declares a public health emergency or because a news event occurs. New outbreak and surveillance signals are incorporated through GIDEON’s standard enrichment and rule assignment on the next production cycle, or sooner if epidemiology initiates an expedited review and publish.

For time-sensitive situations, contact your GIDEON account team regarding override requests and publication timing.

8. How to use and not use these levels

Appropriate uses

• Travel medicine and clinical decision support alongside history, prophylaxis guidelines, and destination-specific advisories
• Teaching and research when GIDEON’s evidence synthesis is cited
• Comparing relative risk across regions within the same disease map

Limitations

• Not a substitute for official national surveillance, legal travel requirements, or individual clinical judgment
• Not real-time outbreak detection. Baseline reflects curated evidence; current weather reflects a rolling environmental window, not case reporting
• Population-level potential, not individual exposure. Levels describe environmental and epidemiological transmission potential for an area, not whether a specific traveler or patient will be exposed or infected
• Local mitigation not modeled. Actual risk can be much lower where individuals or communities use effective prevention (e.g. insecticide-treated nets, indoor residual spraying, air conditioning, personal repellents, vaccination or prophylaxis where applicable). GIDEON’s macro model does not track household- or facility-level vector-control programs or personal behavior
• Not personal risk - behavior, prophylaxis, and microclimate dominate individual outcomes
• Ordinal scale - levels are not incidence rates or probabilities
• Incomplete subnational coverage - many countries rely on country-level or partial administrative coverage until finer maps are added
• Integrators - Partners embedding GIDEON levels in clinical decision support should treat them as one input alongside history, destination advisories, and individual risk factors

9. Illustrative outcomes (composites)

Pedagogical examples only, not live map extracts.

Example A : Baseline only

Example B : Weather lowering (when feature enabled)

Example C : Override vs automated assignment

10. Proprietary information and optional expert review

What this document does not disclose

© 2026 GIDEON Informatics. Proprietary methodology; unauthorized redistribution prohibited.