Evidence workflow
- Define the institutional decision, geography, population, indicators and intended use.
- Identify authoritative, dated sources and document definitions, gaps and limitations.
- Triangulate country evidence with program, stakeholder and implementation context.
- Apply qualified review for health, legal, protection or policy claims as relevant.
- Publish citations, assumptions, update dates, conflict disclosures and correction routes.
Starting reference sources
These international sources provide a baseline for scoping. A delivered analysis requires relevant local and topic-specific sources as well.
- World Health Organization: statement on menstrual health and rights
- World Bank: menstrual health and hygiene
- UNICEF: menstrual hygiene and WASH
- UNFPA: menstruation and human rights
- UNESCO: comprehensive sexuality education
- WHO: ethics and governance of AI for health guidance on large multi-modal models
Visual assets and attribution
Original Mahwari Nexus SVG visualizations are used for the intelligence operating loop, report builder, supply-chain traceability and youth safe companion. Open-source public-health icons are drawn from the Health Icons project for neutral data, privacy, WASH and document visual cues.
Open-source icon source: Health Icons on GitHub and healthicons.org.
Download and export policy
Public static-release downloads are limited to reviewable file formats such as PDF and CSV. Future premium exports may add DOCX, XLSX and image formats for approved report, chart and table outputs. The platform does not ship hidden HTML downloads as final reports.
Dashboard data release
The public dashboard currently renders a global latest-available contextual proxy snapshot imported from the World Bank World Development Indicators API: 1,120 enriched observations across 217 country/economy geographies, excluding regional aggregates. Six priority focus countries also retain 204 dated observations for 2018-2024 trend exploration. These values describe enabling context such as basic water, sanitation, education, poverty and related indicators; they are not direct measures of menstrual health, period poverty, workplace equity or policy performance.
- Official: an observation published through an identified official public dataset.
- Modeled: an official-source observation identified by its publisher as a modeled estimate.
- Curated: a public signal organized for discovery and requiring human validation.
- Prototype index: a Mahwari Nexus analytical framework, not an official statistic.
Each rendered numeric observation records its indicator name, country, region, year, value, source, source URL, update frequency, retrieval date, notes, confidence description, data type and classification. The local data release is structured in /data/global-indicators.json, /data/indicators.json, /data/sources.json, /data/countries.json, /data/country-profiles.json, /data/policy-tracker.json, /data/startups.json, /data/data-quality.json, /data/autonomous-indices.json, /data/index-methodologies.json, /data/period-poverty-risk.json, /data/school-readiness.json, /data/workplace-readiness.json, /data/humanitarian-mhm.json, /data/opportunities.json, /data/reports.json, /data/content-glossary.json, /data/autonomous-os.json and /data/methodology.json.
Autonomous data-quality layer
The preview validation engine detects missing indicators, older latest observations, modeled estimates, unsupported values and source-implementation status. Badges such as official data, modeled estimate, data gap, outdated and needs verification are generated to stop sample or proxy data from looking more certain than it is.
Composite index previews
No proprietary index is represented as an official national score. The autonomous index engine calculates ten Mahwari Nexus methodology previews from available public proxies and curated signals; missing direct inputs are listed and are not imputed.
Period Poverty Risk Index
Preview components: poverty rate plus inverse water, sanitation and female-secondary-enrolment proxy values. Still missing verified product affordability, rural/urban disparity, humanitarian vulnerability and policy-support inputs.
School Readiness Score
Preview components: basic water, sanitation and female-secondary-enrolment contextual proxies. A valid school score still requires toilets/privacy, disposal, CSE, teacher, product and community data.
Workplace Equity Score
Not scored from country proxies. An organizational audit must assess facilities, products, HR policy, flexibility, training and employee-safe feedback.
Policy, Humanitarian, Startup and Implementation
Policy scores express evidence readiness until verified. Humanitarian, product affordability, waste, startup, SRHR and implementation scores are planning previews that require field, legal, clinical or operational review before use.
Autonomous operating-system boundary
Mahwari Nexus automates structure, routing, quality checks, report previews, proposal skeletons and recommendation prompts. It does not autonomously publish reports, diagnose medical issues, issue legal advice, send client proposals or make procurement decisions. All sensitive outputs remain human-reviewable.
Supply-chain transparency pilot
The supply-chain transparency page is a sample demo for QR product passports, batch event trails and future Hyperledger Fabric-style permissioned ledgers. It is not live-chain proof, product certification, procurement due diligence or supplier endorsement. No beneficiary names, student identities or personal health data should be written to traceability records.
Missing data and update cycle
The dashboard displays unavailable observations rather than estimating them silently. Data imports should be refreshed when official series update, then reviewed for definition or vintage changes before publication. National policy, affordability, school, workplace and humanitarian evidence requires qualified human review before it can support recommendations.
AI-assisted outputs
The public dashboard evidence assistant is deterministic: it can answer supported inventory, indicator, country-comparison, ranked-proxy and curated startup-signal questions directly from the published release. It cannot retrieve confidential documents, verify policy claims or issue medical conclusions. A future retrieval-assisted system should only use approved knowledge sources, preserve citations, express confidence and assumptions, label generated material, log review and protect uploaded documents.
Publication and corrections
Reports and major knowledge pages should identify authorship, reviewers, sources, data vintage, limitations, disclosures, update dates and a corrections channel before being represented as authoritative outputs.