# DE Wastewater Surveillance (AMELAG)

---

## 🎯 BEST ESTIMATE

| Metric | Value | Confidence | Last Updated |
|--------|-------|------------|--------------|
| **SARS-CoV-2 viral load (national)** | **~1,554 gc/L** | 90% | 2026-04-15 |
| **SARS-CoV-2 4-week trend** | **-52.8% (↓↓)** | 85% | 2026-04-15 |
| **SARS-CoV-2 % of all-time peak** | **0.3%** | 90% | 2026-04-15 |
| **Influenza A 4-week trend** | **-39.3% (↓↓)** | 85% | 2026-04-15 |
| **RSV B 4-week trend** | **-27.0% (↓↓)** | 85% | 2026-04-15 |

**One-liner:** All five tracked respiratory pathogens declining in German wastewater, SARS-CoV-2 at 0.3% of peak.

**Caveat:** Wastewater surveillance detects population-level viral shedding, not clinical severity; seasonal declines are expected in spring.

---

## Quick Context

Germany's AMELAG network (Abwassersurveillance für die epidemiologische Lagebewertung) monitors respiratory viruses in wastewater from ~67–75 treatment plants covering ~25% of the population. Wastewater data provides an unbiased signal of community infection levels because it captures all shedding individuals regardless of testing behavior. The system tracks SARS-CoV-2, Influenza A/B, and RSV A/B with weekly population-weighted national aggregates.

---

## Datasets

### wastewater-latest.csv
Full time series for all pathogen types (SARS-CoV-2, Influenza A/B, RSV A/B, and combined variants).
- **datum**: Week date (Wednesday of reporting week)
- **typ**: Pathogen type
- **n_sites**: Number of reporting wastewater treatment plants
- **pop_coverage_pct**: Population coverage percentage
- **viruslast_gc_per_l**: Viral load in gene copies per liter
- **viruslast_normalisiert**: Flow-normalized viral load
- **vorhersage_gam**: GAM-smoothed prediction value
- **obere_schranke / untere_schranke**: 95% confidence intervals

### wastewater-summary.csv
Current status summary per primary pathogen with trend analysis.

---

## Methodology Summary

**Approach:** Population-weighted aggregation of wastewater treatment plant viral load measurements, smoothed with Generalized Additive Models (GAM).

**Sources:**
- [AMELAG — RKI / Umweltbundesamt (GitHub)](https://github.com/robert-koch-institut/Abwassersurveillance_AMELAG)
- [Infektionsradar (Dashboard)](https://infektionsradar.rki.de/de/covid/abwasser)
- [Umweltbundesamt PiA-Monitor](https://www.umweltbundesamt.de/amelag)

**Definition Used:** Gene copies per liter (gc/L) of treated wastewater influent, population-weighted national aggregate across reporting sites.

---

## Source Analysis

### Why These Sources?

| Source | Strengths | Weaknesses | Weight Given |
|--------|-----------|------------|--------------|
| **AMELAG GitHub (RKI/UBA)** | Official open data, CC-BY 4.0, machine-readable TSV, weekly updates | ~25% population coverage, 1-2 week lag | High |
| **Infektionsradar Dashboard** | Visual context, additional metrics | No public API, frontend-only | Reference |

### Key Limitations

1. Population coverage is ~25% — not all German cities/regions are equally represented.
2. Lab changes at individual sites can cause discontinuities (flagged in individual site data).
3. Wastewater measures viral shedding, not infections or severity — the relationship between gc/L and clinical burden is non-linear.

---

## Substrate Connection

- **Problems:** PR-00003 (Performance Society Exhaustion — pandemic impact on work/health)
- **Related Datasets:** Bay-Area-COVID-Wastewater (US equivalent)
- **Cross-reference:** DE-Mental-Health (pandemic impact on mental health indicators)

---

## Research Metadata

| Attribute | Value |
|-----------|-------|
| **Research Date** | 2026-04-22 |
| **Researcher** | Sven / PAI |
| **Method** | Automated fetch from AMELAG GitHub TSV |
| **Confidence Level** | 90% |
| **Known Gaps** | Individual site data not included in summary; variant-level data not available in wastewater |

---

## Changelog

| Date | Change | Reason |
|------|--------|--------|
| 2026-04-22 | Initial dataset creation | Implement DE wastewater surveillance from AMELAG |

---

## Full Data

See `wastewater-latest.csv` for the complete time series (905 data points across all pathogen types).
Fetch script: `bun get-de-wastewater` from Substrate root.