tooluniverse-adverse-event-detection

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass.

Audit 0.4.1 · source sha 43be0fac747b · 28 Oct 2026Report an issue or request a re-scan

• Publishermims-harvard GitHub organizationOwner of the repository this skill is read from
• Source foldermims-harvard/tooluniverse/skills/tooluniverse-adverse-event-detectionThe exact folder we audited
• LicenceApache-2.0As the repository states it
• Commit pinned for this verdict43be0fac747bA new audit runs when the source changes
• Repository statusNot archivedLast push 1 day ago
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Detect and analyze adverse drug event signals using FDA FAERS reports, drug labels, and disproportionality statistics (PRR, ROR, IC). Generates quantitative safety signal scores (0-100) with evidence grading. Use for…

---
name: tooluniverse-adverse-event-detection
description: Detect and analyze adverse drug event signals using FDA FAERS reports, drug labels, and disproportionality statistics (PRR, ROR, IC). Generates quantitative safety signal scores (0-100) with evidence grading. Use for post-market surveillance, pharmacovigilance, drug safety assessment, regulatory submissions, and detecting rare AE signals not visible in clinical trials.
disable-model-invocation: true
---

## COMPUTE, DON'T DESCRIBE
When analysis requires computation (statistics, data processing, scoring, enrichment), write and run Python code via Bash. Don't describe what you would do — execute it and report actual results. Use ToolUniverse tools to retrieve data, then Python (pandas, scipy, statsmodels, matplotlib) to analyze it.

# Adverse Drug Event Signal Detection & Analysis

Automated pipeline for detecting, quantifying, and contextualizing adverse drug event signals using FAERS disproportionality analysis, FDA label mining, mechanism-based prediction, and literature evidence. Produces a quantitative Safety Signal Score (0-100) for regulatory and clinical decision-making.

**KEY PRINCIPLES**:
1. **Signal quantification first** - Every adverse event must have PRR/ROR/IC with confidence intervals
2. **Serious events priority** - Deaths, hospitalizations, life-threatening events always analyzed first
3. **Multi-source triangulation** - FAERS + FDA labels + OpenTargets + DrugBank + literature
4. **Context-aware assessment** - Distinguish drug-specific vs class-wide vs confounding signals
5. **Report-first approach** - Create report file FIRST, update progressively
6. **Evidence grading mandatory** - T1 (regulatory/boxed warning) through T4 (computational)
7. **English-first queries** - Always use English drug names in tool calls, respond in user's language

**REASONING STRATEGY — Start Here**:
Start with the signal: What adverse event was reported more than expected? (PRR >= 2.0, N >= 3, lower CI > 1.0 is the threshold). Then ask three questions in order:
1. **Biologically plausible?** Given the drug's mechanism of action and targets, does this adverse event make sense? An off-target kinase inhibitor causing cardiac events is plausible; a topical agent causing systemic toxicity needs more scrutiny. LOOK UP DON'T GUESS — use `OpenTargets_get_drug_mechanisms_of_action_by_chemblId` and `drugbank_get_targets_by_drug_name_or_drugbank_id` to check targets before asserting plausibility.
2. **Timing consistent?** Acute reactions (within hours/days) suggest immune or direct pharmacologic mechanism. Delayed reactions (weeks/months) suggest cumulative toxicity or idiosyncratic response. Check FAERS time-to-onset distribution.
3. **Could confounders explain it?** Patients taking this drug likely have the underlying disease — compare against background rate in that population, not the general population. Class-wide signals (appearing for all drugs in the class) suggest mechanism-based rather than molecule-specific toxicity.

**Causality Assessment — Naranjo Algorithm Reasoning**:
When determining whether an adverse event is drug-caused (not just associated), apply these steps systematically. LOOK UP DON'T GUESS — search FAERS and FDA labels for each criterion:
1. **Prior reports?** Are there previous conclusive reports of this reaction? Check FDA label (`FDA_get_adverse_reactions_by_drug_name`) and literature (`PubMed_search_articles`). Yes = +1.
2. **Temporal relationship?** Did the AE appear after drug administration? Onset within expected pharmacokinetic window (1-5 half-lives) = +2. Use `FAERS_stratify_by_demographics` for time-to-onset data.
3. **Dechallenge?** Did the AE improve when the drug was stopped? Positive dechallenge = +1. Look for rechallenge/dechallenge case reports in literature.
4. **Rechallenge?** Did the AE reappear when the drug was restarted? Positive rechallenge = +2 (strongest single piece of evidence for causality).
5. **Alternative causes?** Could the underlying disease, concomitant drugs, or other factors explain the AE? Check `drugbank_get_drug_interactions_by_drug_name_or_id` for interacting drugs.
6. **Dose-response?** Did the reaction worsen with higher doses or improve with lower doses? Dose-dependent AEs suggest on-target toxicity.
7. **Drug level confirmation?** Was the drug detected in body fluids at toxic concentrations?
- Score: Definite (>=9), Probable (5-8), Possible (1-4), Doubtful (<=0).
- Even without individual patient data, you can estimate causality from aggregate FAERS signals + label evidence + mechanistic plausibility.

**Reference files** (in this directory):
- `PHASE_DETAILS.md` - Detailed tool calls, code examples, and output templates per phase
- `REPORT_TEMPLATE.md` - Full report template and completeness checklist
- `TOOL_REFERENCE.md` - Tool parameter reference and fallback chains
- `QUICK_START.md` - Quick examples and common drug names

---

## When to Use

Apply when user asks:
- "What are the safety signals for [drug]?"
- "Detect adverse events for [drug]"
- "Is [drug] associated with [adverse event]?"
- "What are the FAERS signals for [drug]?"
- "Compare safety of [drug A] vs [drug B] for [adverse event]"
- "What are the serious adverse events for [drug]?"
- "Are there emerging safety signals for [drug]?"
- "Post-market surveillance report for [drug]"
- "Pharmacovigilance signal detection for [drug]"

**Differentiation from tooluniverse-pharmacovigilance**: This skill focuses specifically on **signal detection and quantification** using disproportionality analysis (PRR, ROR, IC) with statistical rigor, produces a quantitative **Safety Signal Score (0-100)**, and performs **comparative safety analysis** across drug classes.

---

## Workflow Overview

```
Phase 0: Input Parsing & Drug Disambiguation
  Parse drug name, resolve to ChEMBL ID, DrugBank ID
  Identify drug class, mechanism, and approved indications
    |
Phase 1: FAERS Adverse Event Profiling
  Top adverse events by freq

Excerpt, first 6,000 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

git clone https://github.com/mims-harvard/tooluniverse
git -C tooluniverse checkout 43be0fac747b

Works in

1. 2026-10-28
   ✓SAFE B 89 / 100

   first audit · source 43be0fac747b

Questions about tooluniverse-adverse-event-detection

• github.com/mims-harvard/tooluniverse/tree/main/skills/tooluniver — the audited source at commit 43be0fac747b and every finding aboveObserved · 28 Oct 2026

Spotted something wrong? Report a correction · Methodology