# faers.mobi

> Isolated FAERS disproportionality (drug × event). Hypothesis, not causation.

Public aggregates from FDA FAERS. Read `n`, `eb05`, and `trend`. No API key.
Live extract through-quarter: `X-FAERS-Through` on GET /signals (also
`faers_through` on `format=profile` / `Data through` on `format=brief`).

## Worked example: tzield × Nausea

Live body of `GET https://faers.mobi/signals?drug=tzield&event=Nausea&format=brief` (cached 2026-10-14). Same markdown you get from that URL.

### RWE brief: tzield × Nausea

Data through 2025Q4.

### Filters
Applied: none.
Rows omitted: none.
Counts below are over the rows that survived these filters. n and EB05 are per reported drug string and are never summed or averaged across strings.

### Profile
153 pairs · 2 drugs (tzield, teplizumab). Novel 119 / known 34 / ? 0.
Trend: 8 Accelerating, 13 Rising, 116 New, 15 Stable, 1 Declining
Top EWMA EB05: Mononucleosis syndrome (309.4)
Top n: Nausea (319)
Novel+rising: 9

### Class
Other antidiabetics (A10XX01). 1 substance in Other antidiabetics: too few for a class-wide comparison (needs 3); no drug-specific claim made

### Series
11 quarters 2023Q2–2025Q4, sum n=319. Last 4: 2025Q1 n=36 eb05=5.01 ewma=7.01; 2025Q2 n=40 eb05=5.71 ewma=7.81; 2025Q3 n=47 eb05=5.38 ewma=7.73; 2025Q4 n=43 eb05=4.82 ewma=7.14.

### Pair
Isolated FAERS pair: tzield × Nausea (exact/prefix/substring/same-substance isolation; fuzzy near-matches omitted). Across flagged quarters this pair has n = 319 observed reports, peak quarterly EB05 = 5.71 and peak EWMA EB05 = 7.81 (EB05 is the 5th percentile of the dual-Gamma GPS/EBGM posterior; EWMA is the smoothed series Trend uses). Trend is Rising from last-4 vs prior-4 mean EWMA EB05. Novelty is known (the event term was matched in the cached FDA label text). This is a statistical hypothesis from spontaneous reports, not evidence that the drug causes the event.

## Build a brief URL for any drug or event

Replace the names in `https://faers.mobi/signals?drug=<drug>&event=<event>&format=brief` (URL-encode spaces as `%20`). Use only `drug=` for a drug profile or only `event=` for an event profile. The response is markdown like the section above, so paste that URL into ChatGPT, Claude, or Perplexity and ask about it. No API key.

## Docs

- [API.md](https://faers.mobi/API.md): human REST docs for GET /signals
- [Ranking](https://faers.mobi/signals?drug=tzield&indication=hide&low_info=hide&sort=eb05&limit=5): for top-signal asks add `indication=hide&low_info=hide` (rows carry both booleans; default order unchanged) or use format=profile / brief
- [OpenAPI](https://faers.mobi/openapi.json): OpenAPI 3 schema for GET /signals (json / csv / series / profile / class / brief / pdf)
- [ChatGPT Actions](https://faers.mobi/chatgpt.md): paste-ready Custom GPT import of that schema
- [Signals API](https://faers.mobi/signals?drug=tzield&event=Nausea&format=brief): example pair brief
- [Event profile](https://faers.mobi/signals?event=Nausea&format=profile): compact JSON for an isolated event (top drugs)
- [Event brief](https://faers.mobi/signals?event=Nausea&format=brief): markdown RWE brief from that event profile
- [Static example brief](https://faers.mobi/examples/tzield-nausea-brief.txt): the tzield × Nausea brief as a plain file, no query string (refreshed nightly)
- [RWE brief PDF](https://faers.mobi/signals?drug=tzield&event=Nausea&format=pdf): same brief as `format=brief`, Typst PDF (`drug=` or `event=`)
- [Event PDF](https://faers.mobi/signals?event=Nausea&format=pdf): event-only brief as PDF
- [Human UI](https://faers.mobi/): search and selected-pair series table

## Optional

- [Pipeline](https://globalpatientsafety.com/pipeline?sponsor=Sanofi&format=brief): ClinicalTrials.gov studies for one lead sponsor. role=any includes collaborators. Tool get_pipeline on this MCP server. Registry facts, not a FAERS signal.
- [VAERS](https://vaers.globalpatientsafety.com/signals?vaccine=COVID19&event=Myocarditis&format=brief): vaccine reporting signals. Tools get_vaers, vaers_search_vaccine, vaers_get_signal, and vaers_get_profile on this MCP server. Hypothesis, not causation.
- [MCP](https://faers.mobi/mcp.md): hosted Streamable HTTP at https://faers.mobi/mcp (stdio: /mcp/server.py). Includes get_pipeline, get_vaers, vaers_search_vaccine, vaers_get_signal, and vaers_get_profile.
- Perplexity: Settings → Connectors → Add custom connector, URL https://faers.mobi/mcp (steps on /mcp.md)
- [Label status](https://faers.mobi/signals?drug=tzield&event=Nausea&format=label): cached FDA-label Novel for one pair

Ingredient grouping, product type and medication errors (2026-09-27)
- Every /signals row carries reported_drug (the raw FAERS string), ingredient
  (the active it resolves to), product_type (single | combination |
  trial_uncertain) and medication_error (true for use/supply PTs).
- product_type=single keeps only single-ingredient rows; medication_error=hide
  drops medication-error and product-issue rows. Both are opt-in VIEWS, never
  merges: unfiltered output is unchanged and every row carries its flags.
- Novelty is looked up on the ingredient, not the raw string.
- n and EB05 are per reported string and are NEVER summed or averaged across
  strings. Ingredient-level recomputation is a separate offline step.
- format=brief and format=pdf name every filter applied and the rows omitted
  by reason. format=class counts distinct ingredients, excluding the focal
  drug's own brands.