tooluniverse-drug-drug-interaction

Drug-Drug Interaction Prediction Skill

Comprehensive DDI risk assessment with evidence-based clinical guidance.

---

Quick Start

Example 1: Two-Drug Analysis

User: "Analyze drug interactions between warfarin and amoxicillin"

Agent: Creates `DDI_risk_report_warfarin_amoxicillin.md` with:
- Bidirectional mechanism analysis
- Risk score: 55/100 (Moderate)
- Management: Monitor INR at day 3-5
- Evidence: ★★★ (FDA label + clinical studies)

Example 2: Polypharmacy (5+ drugs)

User: "Assess interactions for: warfarin, lisinopril, metoprolol, omeprazole, amlodipine, furosemide"

Agent: Creates `DDI_risk_report_polypharmacy.md` with:
- 15 pairwise interactions analyzed
- DDI matrix showing all combinations
- Polypharmacy risk score: 38/100
- Top 3 priority interactions with management

Example 3: Alternative Drug Recommendations

User: "Which statin is safest with diltiazem and erythromycin?"

Agent: Analyzes all statins and recommends:
- PREFERRED: Pravastatin or rosuvastatin (no CYP3A4 interaction)
- AVOID: Simvastatin, lovastatin (contraindicated)
- Evidence table with risk scores for each statin

---

Core Features

1. Bidirectional Analysis

Always analyzes both directions:

• Drug A → Drug B (A affects B)
• Drug B → Drug A (B affects A)

Effects are often asymmetric (e.g., ketoconazole affects simvastatin, but not vice versa).

2. Multi-Dimensional Risk Scoring (0-100)

Score = Mechanism(30) + Evidence(25) + Clinical Impact(25) + Prevalence(10) + Reversibility(10)

80-100: Critical (contraindicated)
60-79:  High (avoid if possible)
40-59:  Moderate (monitor/adjust)
20-39:  Low (minimal action)
0-19:   Negligible

3. Evidence Grading

• ★★★ = FDA label, RCT, clinical guidelines
• ★★☆ = Clinical studies, PK studies, case series
• ★☆☆ = Case reports, theoretical mechanisms
• ☆☆☆ = Computational predictions only

4. Mechanism Analysis

Pharmacokinetic (PK):

• CYP450 enzyme interactions (CYP3A4, 2D6, 2C9, 2C19, 1A2)
• Drug transporter interactions (P-gp, OATs, OCTs, BCRP, OATPs)
• Protein binding displacement

Pharmacodynamic (PD):

• Additive/synergistic effects (e.g., two anticoagulants)
• Antagonistic effects (e.g., beta-blocker + beta-agonist)
• QTc prolongation (additive cardiac risk)
• CNS depression (additive sedation)
• Serotonin syndrome risk

5. Clinical Management Strategies

For each DDI, provides:

1. Avoid combination (if Major/contraindicated)
2. Alternative drug recommendations
3. Dose adjustments
4. Timing separation strategies
5. Monitoring parameters (labs, vital signs, symptoms)
6. Patient counseling points

---

When to Use This Skill

✅ Use for:

• Two-drug interaction analysis
• Polypharmacy risk assessment (3+ drugs)
• Alternative drug recommendations (lower DDI risk)
• QTc prolongation risk (multiple drugs)
• Clinical decision support for prescribing
• Patient safety reviews
• Medication reconciliation

❌ Don't use for:

• Single drug safety profile → use tooluniverse-drug-research
• Drug mechanism only → use tooluniverse-drug-research
• Pharmacogenomics → use PGx-specific analysis
• Disease-drug associations → use disease-focused skills

---

Key Principles

Report-First Approach

1. Create report file FIRST with all section headers
2. Populate progressively as data is gathered
3. User sees complete report, not search process

Patient Safety Focus

• Actionable clinical guidance, not just theory
• Evidence-based recommendations
• Alternative drugs when DDI risk is high
• Clear monitoring parameters
• Patient-friendly counseling language

Completeness Requirement

• All sections must exist
• Explicit "No interaction found" when appropriate
• Document tool failures
• Cite data sources throughout

---

Tools Used

Analysis Path	Primary Tools
Drug ID	rxnorm_get_drugs_by_name, DailyMed_search_spls, PubChem_get_CID_by_compound_name
CYP interactions	ADMETAI_predict_CYP_interactions, DailyMed_get_spl_sections_by_setid
FDA warnings	DailyMed_get_spl_sections_by_setid (contraindications, boxed_warning, drug_interactions)
Clinical evidence	PubMed_search_articles, search_clinical_trials
Post-marketing	FAERS_count_reactions_by_drug_event
Alternatives	DGIdb_get_drug_gene_interactions, DailyMed_search_spls

---

Output Format

Report Structure

DDI_risk_report_[DRUG1]_[DRUG2].md

0. Drug Identification
1. Interaction Mechanisms (CYP, Transporters, PD)
2. FDA Label & Regulatory Warnings
3. Clinical Evidence & Literature
4. Post-Marketing Surveillance (FAERS)
5. Severity Classification & Risk Scoring
6. Clinical Management Strategies
7. Summary & Clinical Action Plan

Polypharmacy Report

DDI_risk_report_polypharmacy.md

- DDI matrix (all pairwise combinations)
- Polypharmacy risk score
- Top priority interactions
- Cumulative monitoring plan
- Simplification opportunities

---

Real-World Examples

See EXAMPLES.md for 7 detailed scenarios:

1. Warfarin + Amoxicillin - CYP/gut flora interaction
2. Simvastatin + Ketoconazole - Major contraindicated DDI
3. Polypharmacy (6 drugs) - Complex elderly regimen
4. QTc prolongation - Multiple psychiatric drugs
5. Alternative recommendations - Safer statin selection
6. Drug-food interaction - Tacrolimus + grapefruit
7. Timing separation - Levothyroxine + calcium

---

Success Criteria

Before finalizing DDI report:

✅ All drug names resolved to standard identifiers ✅ Bidirectional analysis completed (A→B and B→A) ✅ All mechanism types assessed (CYP, transporters, PD) ✅ FDA label warnings extracted ✅ Clinical literature searched ✅ Evidence grades assigned (★★★, ★★☆, ★☆☆) ✅ Risk score calculated (0-100) ✅ Severity classified (Major/Moderate/Minor) ✅ Primary management recommendation provided ✅ Alternative drugs suggested (if Major DDI) ✅ Monitoring parameters defined ✅ Patient counseling points included ✅ All sections completed (no [Analyzing...] placeholders) ✅ Data sources cited throughout

When all criteria met → Ready for Clinical Use

---

Quick Reference

Severity Classification

• Major: High risk, life-threatening. Avoid combination
• Moderate: Significant risk. Monitor closely, dose adjust
• Minor: Limited risk. Minimal action needed

Risk Score Interpretation

• 80-100: Critical - contraindicated
• 60-79: High - avoid if alternatives exist
• 40-59: Moderate - monitor and adjust
• 20-39: Low - minimal precautions
• 0-19: Negligible

Management Hierarchy

1. Avoid combination (Major DDI)
2. Use alternative drug
3. Dose adjustment
4. Timing separation
5. Intensive monitoring
6. Patient education

---

Technical Details

Evidence Quality Metrics

Reports include:

• Distribution of evidence grades (% ★★★, ★★☆, ★☆☆)
• Source coverage (FDA labels, PubMed, FAERS)
• Data recency and limitations

Limitations

• FAERS does not support direct co-medication queries (requires manual review)
• Some theoretical interactions lack clinical validation
• Alternative drug recommendations limited to same therapeutic class
• Non-US regulatory data (EMA, PMDA) not available via public APIs

---

For Developers

Skill Implementation

See SKILL.md for:

• Complete implementation guide
• Tool call sequences
• Error handling patterns
• Bidirectional analysis algorithms
• Risk scoring formulas
• Alternative drug selection logic

Testing

Validate skill with:

1. Two-drug pairs (10+ examples)
2. Polypharmacy (3 scenarios, 5-8 drugs each)
3. Edge cases (no interaction, contraindicated, food-drug)
4. Bidirectional asymmetry (e.g., A→B but not B→A)

---

Clinical Disclaimer

This skill provides decision support information based on FDA labels, clinical literature, and pharmacology databases. It is NOT a substitute for clinical judgment, patient-specific assessment, or consultation with a pharmacist. Always verify critical DDI information with authoritative sources and consider individual patient factors (age, renal/hepatic function, comorbidities, concomitant medications beyond those analyzed).

---

Version Information

Skill Version: 1.0 Last Updated: 2026-02-09 ToolUniverse Compatibility: v0.3.0+ Key Dependencies: DailyMed, PubMed, ADMET-AI, FAERS, RxNorm

---

For questions or feedback, see main ToolUniverse documentation.