mmrech/icskg-br-processed

---
license: cc-by-4.0
language:
- pt
- en
pretty_name: ICSKG-BR Processed Data
size_categories:
- 10K<n<100K
tags:
- global-surgery
- urban-development
- brazil
- epidemiology
- lcogs
- cimi
- panel-data
- datasus
---

# ICSKG-BR: Index of Cities' Smartness & Knowledge for Global Surgery — Brazil

Processed data layer for the ICSKG-BR longitudinal ecological panel study.
Cross-references the IESE Cities in Motion Index (CIMI) urban development
framework against Lancet Commission on Global Surgery (LCoGS) indicators
across all 5,570 Brazilian municipalities (2015-2023).

**This dataset is currently private during the BMJ Global Health pre-submission
window. It will become public (CC-BY-4.0) upon publication.**

## Provenance

- Source code: https://github.com/matheus-rech/ICSKG
- Dataset version: `v0.1.0`
- Schema version: `1.1`
- Generated at: `2026-04-07T15:59:23Z`
- Source DuckDB: `icskg_br_export.duckdb` (sha256 `473d5ad523dab966`)
- Compression: `zstd` level `6`

## Cookbook scope (v0.1.0)

This release implements the **LCoGS-side** of the ICSKG-BR Technical
Cookbook v1.0 (March 2026). The canonical `panel/municipal_health.parquet`
is derived at publish time from the source tables via the cookbook §5
merge pipeline (50,130 rows = 5,570 mun × 9 years, 49 columns).

**Cookbook §3 sources INCLUDED in v0.1.0:**

- IBGE population master (§3.3)
- DATASUS SIH surgical aggregates (§3.1)
- DATASUS CNES SAO + bellwether (§3.2)
- IBGE SIDRA GDP (§3.3)
- Atlas Brasil HDI (§3.4)
- FIRJAN IFGF (§3.5)
- ANS TABNET insurance (§3.9)
- Census 2022 sanitation (§3.8 partial)
- Mobility (vehicle fleet)

**DEFERRED to v0.2.0** (extractors not yet built — see project phase 12):

- ANATEL broadband (§3.6) — needed for CUDS Technology dimension
- RAIS employment (§3.7) — needed for CUDS Economy/Workforce dimension
- SNIS sanitation proper (§3.8) — Census 2022 is a partial proxy
- SIOPS health spending (§3.10) — needed for CUDS Governance dimension
- International comparators (§3.11) — WHO/World Bank/UNDP
- CUDS composite + dimension scores (§6) — blocked by missing sources above

The v0.2.0 release will add the missing 5 extractors and the cookbook §6
CUDS composite (PCA weighting + geometric mean aggregation), and will
move the `municipal_health` derivation into the build pipeline so it's
stored as a base table in the source DuckDB instead of being computed
at publish time.

## Totals

| Metric | Value |
|---|---|
| Tables | 23 |
| Total rows | 20,037,428 |
| Compressed size | 232.80 MB |

## Tables by namespace

### `panel/`

| Table | Rows | Columns | Size |
|---|---|---|---|
| `municipal_health` | 50,130 | 49 | 2601.98 KB |

### `lcogs/`

| Table | Rows | Columns | Size |
|---|---|---|---|
| `lcogs1_access` | 5,571 | 6 | 43.50 KB |
| `v_lcogs_latest` | 5,570 | 30 | 367.28 KB |
| `v_lcogs_panel` | 50,130 | 30 | 2442.88 KB |
| `v_lcogs_summary` | 49,717 | 26 | 1394.48 KB |

### `source_tables/`

| Table | Rows | Columns | Size |
|---|---|---|---|
| `ans_cobertura` | 5,594 | 6 | 60.58 KB |
| `bellwether_hospitals` | 765 | 7 | 23.84 KB |
| `censo2022_saneamento` | 5,570 | 4 | 47.88 KB |
| `cnes_professionals_raw` | 218,461 | 6 | 129.47 KB |
| `frota_veiculos` | 5,571 | 3 | 44.96 KB |
| `gdp` | 50,130 | 4 | 418.16 KB |
| `idhm` | 5,564 | 7 | 53.21 KB |
| `idhm_historical` | 16,692 | 7 | 113.26 KB |
| `ifgf` | 50,112 | 7 | 581.19 KB |
| `mobility` | 5,535 | 3 | 35.73 KB |
| `municipalities` | 5,571 | 12 | 154.55 KB |
| `national_indicators` | 51 | 5 | 3.08 KB |
| `pipeline_log` | 490 | 9 | 9.07 KB |
| `population` | 50,130 | 3 | 207.30 KB |
| `sao_workforce` | 49,753 | 8 | 383.22 KB |
| `sih_municipal` | 50,170 | 8 | 442.42 KB |
| `sih_raw` | 19,355,506 | 25 | 228820.14 KB |
| `v_sao_annual` | 645 | 6 | 5.82 KB |

## Loading

```python
from huggingface_hub import snapshot_download
import duckdb

local = snapshot_download(
repo_id='mmrech/icskg-br-processed',
repo_type='dataset',
revision='v0.1.0',
allow_patterns=['**/*.parquet', 'manifest.json'],
)

# Read the main panel directly:
panel = duckdb.sql(f"SELECT * FROM '{local}/panel/municipal_health.parquet'").df()
```

## Citation

If you use this dataset, please cite the project:

```
Rech, Matheus M. (2026). ICSKG-BR: Index of Cities' Smartness & Knowledge
for Global Surgery — Brazil. https://github.com/matheus-rech/ICSKG
```

Production Zenodo DOI will be minted at BMJ acceptance.

## License

Source data is public under Brazilian Lei de Acesso à Informação
(Law 12.527/2011). This processed dataset is released under CC-BY-4.0
after BMJ Global Health publication.