Canadian Healthy Aging Policy Observatory

A research infrastructure for monitoring, quantifying, and evaluating the effects of aging-related public policy across Canadian governments.

🌐 Live site: acp.icareu.cc

Part of the Healthy Aging Intelligence Lab (HAIL) — using AI, data science, and policy analysis to help Canada build a fairer, more efficient, and more sustainable system of support for an aging population.

Status: Phase 5 — Researcher dashboard + SaaS seams. All five modules have a working v1, now tied together by a cross-cutting dashboard at /: a KPI strip (policies · indicators · observations · HAPI scores · findings · references), a per-jurisdiction HAPI snapshot (overall + care-access composite), recent policies and recent findings (Association/Causal tagged), the live jurisdiction tree, and a shared top nav across every module. HAPI v1 now spans all six domains — Health (StatCan life expectancy at 65, Table 13-10-0389), Independence (CCHS functional health, Table 13-10-0966), Social Participation (CCHS community belonging, Table 13-10-0096), Financial Security (StatCan seniors' low-income rate, Table 11-10-0135), Care Access (CIHI home care), and Digital Inclusion (StatCan seniors' internet use, Table 22-10-0135) — so overall is a genuine six-domain composite, not a restatement of one domain. The composite is weighted with theory-anchored expert tiers (Health + Care Access tier 1; Financial Security + Independence tier 2; Social Participation + Digital Inclusion tier 3) and sensitivity-tested against equal and empirical schemes (hapi weights) — the headline scores are robust to the weighting choice. See pipeline/hapi_pipeline/indicators/weighting.py. Under the hood the platform is now multi-tenant-ready: a single access seam (apps/web/lib/access.ts) resolves "who is asking" and every org-scoped read routes through it, backed by nullable org_id columns (db/migrations/0004) — a no-op single-tenant today, switch-on-able for SaaS (Stage 2/3) without a schema rewrite. The earlier phases stand: Policy Analytics (Tier-1 Association trends + a worked interrupted time series, Causal(ITS), with assumptions/limitations) and the AI Research Assistant (topic → cited evidence pack + Claude-drafted review). The web app surfaces everything: / · /policies · /hapi · /data · /analytics · /assistant. See docs/11-implementation-roadmap.md.

⚠️ This environment can't reach the live source domains, so connectors run against vendored sample fixtures by default (realistic but not official statistics; provenance recorded as fixture:…). Run hapi ingest --live where the network allows.

---

Why an "Observatory," not a website

A policy website collects and displays documents. An observatory does something categorically different: it continuously observes a system, measures it against a consistent yardstick, and reports change over time. That distinction is the whole point of this project.

A policy website	This observatory
Stores policy text	Tracks each policy's lifecycle, budget, target population, and KPIs over time
Uses whatever metrics governments publish	Maintains an independent, transparent index (HAPI) with documented methodology
Answers "what policies exist?"	Answers "did policy X move outcome Y — and how confidently can we say so?"
A static reference	A reproducible, versioned, citable research instrument

The five modules

#	Module	What it does
①	Policy Library	A jurisdiction-aware, time-axis catalog of aging policies (Canada → Federal / Nova Scotia), each record carrying release date, department, full text, AI summary, budget, target population, KPIs, and lifecycle.
②	Data Hub	A versioned, lineage-tracked ingestion layer over Canada's open data: StatCan, Health Infobase, CIHI, Nova Scotia Open Data, Census, and more.
③	Indicators / HAPI	The Healthy Aging Policy Index — an independent index across six domains (Health, Independence, Social Participation, Financial Security, Care Access, Digital Inclusion) with documented data sourcing and scoring.
④	Policy Analytics	Association analysis over open data — with explicit, rigorous separation of correlation from causation (ITS, DiD, synthetic control).
⑤	AI Research Assistant	A retrieval-augmented agent over the policy corpus + literature that finds policies, papers, data, and trends, and drafts cited literature reviews.

Documentation index

Doc	Contents
00-vision.md	HAIL Lab mission & vision; the "observatory ≠ website" thesis; value proposition
01-platform-overview.md	How the five modules fit together; system context diagram; user roles
02-architecture.md	Tech stack & rationale; monorepo layout; deployment & SaaS path; security/privacy
03-data-model.md	Core schema (Jurisdiction / Policy / Indicator / DataSource / Observation); ER diagram
04-module-policy-library.md	Policy Library design
05-module-data-hub.md	Data Hub & ETL pipeline design
06-module-indicators-hapi.md	HAPI indicator system & scoring methodology
07-module-policy-analytics.md	Analytics & causal-inference rigor
08-module-ai-research-assistant.md	AI research assistant design
09-research-roadmap.md	The four-paper research arc & research-asset ledger
10-data-sources-catalog.md	Concrete NS + Federal data sources (URLs, access, licence, cadence)
11-implementation-roadmap.md	Phased build plan & milestones

Tech stack

• Frontend / dashboards: Next.js (TypeScript)
• Data pipeline / analytics: Python (ingestion, ETL, statistics)
• Storage: PostgreSQL
• AI layer: Claude API (policy summarization, the research assistant, HAPI scoring assistance)

See docs/02-architecture.md for the full rationale and monorepo layout.

Repository layout

apps/web/            Next.js (TypeScript) — UI/dashboards; reads the data model
pipeline/            Python — ingestion, ETL, HAPI indicators, analytics (skeleton)
db/                  schema, versioned migrations, seed, and a psql migration runner
packages/contracts/  shared, language-neutral enum contracts (TS + Python)
docs/                the v1 design whitepaper
docker-compose.yml   local Postgres

Getting started (Phase 1)

cp .env.example .env
docker compose up -d db                 # local Postgres
npm run db:migrate -- --seed            # apply schema + seed Canada → Federal / NS
                                         # (or: bash db/migrate.sh --seed)

cd pipeline && pip install -r requirements.txt && \
  python -m hapi_pipeline.cli ingest && \
  python -m hapi_pipeline.cli observations   # load Data Hub + show lineage
cd ..

npm install
echo "DATABASE_URL=postgresql://hapi:hapi_dev_password@localhost:5432/hapi" > apps/web/.env.local
npm run dev                              # http://localhost:3000 — tree + /data lineage page

Per-area setup: db/README.md, apps/web/README.md, pipeline/README.md. To pull real data (live connectors), automate ingestion, and deploy the web app to Google Cloud Run, see RUNBOOK.md (§B live data · §C scheduled ingest · §D Cloud Run).

Deploy the web tier in one command (Cloud Build builds the Dockerfile):

gcloud run deploy aging-policy-lab --source . --region us-east1 \
  --allow-unauthenticated --update-secrets DATABASE_URL=DATABASE_URL:latest

Scope of v1

• Geography: Nova Scotia + Federal (a replicable template; pan-Canadian expansion later).
• Now live: the design whitepaper, the Phase 1 scaffold (monorepo, schema, seed, web read), and the Phase 2 Data Hub (lineage-tracked ingestion + /data view). HAPI scoring comes in Phase 3 (docs/11).

License

Documentation is intended to be released under CC-BY-4.0 (see LICENSE). The licence for future code will be decided when the codebase is created.

A note on rigor

This platform analyzes public data to surface associations and to support evidence-based evaluation. Association is not causation. Causal claims require careful quasi-experimental designs (interrupted time series, difference-in-differences, synthetic control), each with stated assumptions and limitations. The observatory is built to make that distinction explicit at every step — see docs/07-module-policy-analytics.md.

---

中文概览

aging-policy-lab 不是一个普通的政策网站,而是一座研究基础设施(Research Infrastructure)。它的核心不是"收集政策",而是持续监测、量化评估、预测加拿大各级政府老龄化政策的效果。

平台定位为 Canadian Healthy Aging Policy Observatory(加拿大健康老龄化政策观察平台),隶属于一个十年愿景的实验室 Healthy Aging Intelligence Lab(HAIL)。它同时承担三重角色:

• 研究平台 —— 持续产出论文与数据,而非一次性课题;
• 创业项目 —— 未来可发展为 SaaS 或研究服务;
• 研究资产库 —— 不断积累数据库、指标体系、AI Agent、文献知识库、政策知识图谱与可视化看板。

第一版(v1)先聚焦 Nova Scotia + 联邦(Federal) 作为可复制的模板,未来扩展至全加拿大。

平台由五个核心模块组成:① Policy Library(政策库)② Data Hub(数据中心)③ Indicators / HAPI(指标体系)④ Policy Analytics(政策分析)⑤ AI Research Assistant(AI 研究助手)。在线访问:https://acp.icareu.cc。