This source did not publish a separate summary. Review SKILL.md before using the skill.
SKILL.md
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:
Signal quantification first - Every adverse event must have PRR/ROR/IC with confidence intervals
Evidence grading mandatory - T1 (regulatory/boxed warning) through T4 (computational)
English-first queries - Always use English drug names in tool calls, respond in user's language
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]"
"What is the disproportionality analysis for [drug] and [event]?"
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 , and performs across drug classes. The pharmacovigilance skill provides broader safety profiling without the same depth of signal detection metrics.
Safety Signal Score (0-100)
comparative safety analysis
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 frequency
Seriousness and outcome distributions
Demographics (age, sex, country)
|
Phase 2: Disproportionality Analysis (Signal Detection)
Calculate PRR, ROR, IC with 95% CI for each AE
Apply signal detection criteria
Classify signal strength (Strong/Moderate/Weak/None)
|
Phase 3: FDA Label Safety Information
Boxed warnings, contraindications
Warnings and precautions, adverse reactions
Drug interactions, special populations
|
Phase 4: Mechanism-Based Adverse Event Context
Target-based AE prediction (OpenTargets safety)
Off-target effects, ADMET predictions
Drug class effects comparison
|
Phase 5: Comparative Safety Analysis
Compare to drugs in same class
Identify unique vs class-wide signals
Head-to-head disproportionality comparison
|
Phase 6: Drug-Drug Interactions & Risk Factors
Known DDIs causing AEs
Pharmacogenomic risk factors (PharmGKB)
FDA PGx biomarkers
|
Phase 7: Literature Evidence
PubMed safety studies, case reports
OpenAlex citation analysis
Preprint emerging signals (EuropePMC)
|
Phase 8: Risk Assessment & Safety Signal Score
Calculate Safety Signal Score (0-100)
Evidence grading (T1-T4) for each signal
Clinical significance assessment
|
Phase 9: Report Synthesis & Recommendations
Monitoring recommendations
Risk mitigation strategies
Completeness checklist
IMPORTANT: FDA label tools return {error: {code: "NOT_FOUND"}} when a section does not exist. This is NORMAL for many drugs - for example, most drugs do NOT have boxed warnings. Always check for this pattern:
# Check if boxed warning exists
if isinstance(boxed, dict) and 'error' in boxed:
boxed_warning_text = "None (no boxed warning for this drug)"
else:
boxed_warning_text = boxed['results'][0].get('boxed_warning', ['None'])[0]
3.3 Output for Report
## 4. FDA Label Safety Information
### 4.1 Boxed Warning
None
### 4.2 Contraindications
- Acute liver failure or decompensated cirrhosis
- Hypersensitivity to atorvastatin (includes anaphylaxis, angioedema, SJS, TEN)
### 4.3 Warnings and Precautions
| Warning | Clinical Relevance |
|---------|-------------------|
| Myopathy/Rhabdomyolysis | Risk with CYP3A4 inhibitors, high doses |
| Immune-Mediated Necrotizing Myopathy | Rare autoimmune myopathy |
| Hepatic Dysfunction | Monitor LFTs |
| Increased HbA1c/Glucose | Diabetes risk |
### 4.4 Drug Interactions (from label)
| Interacting Drug | Mechanism | Clinical Action |
|-----------------|-----------|-----------------|
| Cyclosporine | Increased exposure | Avoid combination |
| CYP3A4 inhibitors | Increased atorvastatin levels | Use lowest dose |
| Gemfibrozil | Increased myopathy risk | Avoid |
### 4.5 Special Populations
- **Pregnancy**: Contraindicated
- **Geriatric**: No dose adjustment needed
- **Pediatric**: Approved for heterozygous FH ages 10+
*Source: FDA drug labels via FDA_get_contraindications_by_drug_name, FDA_get_warnings_by_drug_name*
Phase 4: Mechanism-Based Adverse Event Context
4.1 Target Safety Profile
# Get target safety data from OpenTargets
# First get target ensembl ID from MOA result
target_id = "ENSG00000113161" # HMGCR from Phase 0
safety = tu.tools.OpenTargets_get_target_safety_profile_by_ensemblID(ensemblId=target_id)
# Response: {data: {target: {id: "...", approvedSymbol: "HMGCR",
# safetyLiabilities: [{event: "Decrease, Fertility", eventId: "...",
# effects: [{direction: "Inhibition/Decrease/Downregulation"}],
# studies: [{type: "cell-based"}], datasource: "AOP-Wiki"}]}}}
# Get OpenTargets adverse events (uses FAERS data)
ot_aes = tu.tools.OpenTargets_get_drug_adverse_events_by_chemblId(chemblId="CHEMBL1487")
# Response: {data: {drug: {adverseEvents: {count: 13, criticalValue: 513.67,
# rows: [{name: "myalgia", meddraCode: "10028411", count: 4126, logLR: 6067.33}, ...]}}}}
4.2 ADMET Predictions (if SMILES available)
# Get SMILES from DrugBank/PharmGKB
smiles = "CC(C)C1=C(C(=C(N1CC[C@H](C[C@H](CC(=O)O)O)O)C2=CC=C(C=C2)F)C3=CC=CC=C3)C(=O)NC4=CC=CC=C4"
# Toxicity predictions
toxicity = tu.tools.ADMETAI_predict_toxicity(smiles=[smiles])
# Response: predictions for hepatotoxicity, cardiotoxicity, etc.
# CYP interaction predictions
cyp = tu.tools.ADMETAI_predict_CYP_interactions(smiles=[smiles])
# Response: CYP inhibition/substrate predictions
4.3 Drug Warnings from OpenTargets
# Drug warnings (withdrawals, safety warnings)
warnings = tu.tools.OpenTargets_get_drug_warnings_by_chemblId(chemblId="CHEMBL1487")
# Response: {data: {drug: {id: "CHEMBL1487", name: "ATORVASTATIN"}}}
# Note: Empty if no warnings exist
The Safety Signal Score quantifies overall drug safety concern on a 0-100 scale (higher = more concern).
Component 1: FAERS Signal Strength (0-35 points)
If any signal has PRR >= 5 AND ROR lower CI >= 3: 35 points
If any signal has PRR 3-5 AND ROR lower CI 2-3: 20 points
If any signal has PRR 2-3 AND ROR lower CI 1-2: 10 points
If no signals detected: 0 points
Component 2: Serious Adverse Events (0-30 points)
Deaths reported with high count (>100): 30 points
Deaths reported with low count (1-100): 25 points
Life-threatening events: 20 points
Hospitalizations only: 15 points
Non-serious only: 0 points
Component 3: FDA Label Warnings (0-25 points)
Boxed warning present: 25 points
Drug withdrawn or restricted: 25 points
Contraindications present: 15 points
Warnings and precautions: 10 points
Adverse reactions only: 5 points
No label warnings: 0 points
Component 4: Literature Evidence (0-10 points)
Meta-analyses confirming safety signals: 10 points
Multiple RCTs with safety concerns: 7 points
Case reports/case series: 4 points
No published safety concerns: 0 points
Pattern 1: Full Safety Signal Profile for a Single Drug
Use all phases (0-9) for comprehensive report. Best for regulatory submissions, safety reviews.
Pattern 2: Specific Adverse Event Investigation
Focus on Phases 0, 2, 3, 7. User asks "Does [drug] cause [event]?" - calculate disproportionality for that specific event, check label, search literature.
Pattern 3: Drug Class Comparison
Focus on Phases 0, 2, 5. Compare 3-5 drugs in same class for a specific adverse event using FAERS_compare_drugs.
Pattern 4: Emerging Signal Detection
Focus on Phases 1, 2, 7. Screen top 20+ FAERS events for signals, identify any not in FDA label (Phase 3), search recent literature for confirmation.
Pattern 5: Pharmacogenomic Risk Assessment
Focus on Phases 0, 6. Identify genetic risk factors for adverse events using PharmGKB and FDA PGx biomarkers.
Pattern 6: Pre-Approval Safety Assessment
Focus on Phases 4, 7. Use ADMET predictions and target safety profiles when FAERS data is limited (new drugs).
Edge Cases
Drug with No FAERS Reports
Skip Phases 1-2
Rely on FDA label (Phase 3), mechanism predictions (Phase 4), and literature (Phase 7)
Safety Signal Score will be lower due to lack of signal detection data
Generic vs Brand Name
Always try both names in FAERS queries (FAERS uses brand names sometimes)
Use OpenTargets_get_drug_chembId_by_generic_name to resolve to standard identifier
Use FDA_get_brand_name_generic_name for name cross-reference
Drug Combinations
Use FAERS_search_reports_by_drug_combination for polypharmacy analysis
Distinguish combination AEs from individual drug AEs
Use FAERS_count_additive_adverse_reactions for aggregate class analysis
Confounding by Indication
Compare AE profile to the disease being treated
Example: "Death" reports for chemotherapy drugs may reflect disease progression
Always note this limitation in the report
Drugs with Boxed Warnings
Score component automatically 25/25 for label warnings
Prioritize boxed warning events in disproportionality analysis
Cross-reference boxed warning with FAERS signal strength
