Skip to content

tooluniverse-clinical-trial-design

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History
 
 

tooluniverse-clinical-trial-design

README.md

Clinical Trial Design Feasibility Skill

Strategic clinical trial design feasibility assessment for early-phase trials (Phase 1/2 emphasis).

What This Skill Does

Systematically evaluates clinical trial feasibility across 6 research dimensions:

  1. Patient Population Sizing - Prevalence, biomarker rates, enrollment projections
  2. Biomarker Strategy - Testing availability, turnaround time, CDx landscape
  3. Comparator Selection - SOC analysis, historical controls, single-arm vs. randomized
  4. Endpoint Selection - Regulatory precedents, measurement feasibility
  5. Safety Monitoring - Mechanism-based toxicities, monitoring plans, DLT definitions
  6. Regulatory Pathway - 505(b)(1), breakthrough therapy, orphan designation

Output: Comprehensive feasibility report with quantitative feasibility score (0-100), enrollment timelines, and go/no-go recommendations.

When to Use This Skill

Use this skill when you need to:

  • Plan Phase 1/2 trials (early development focus)
  • Assess enrollment feasibility for biomarker-selected trials
  • Design basket or umbrella trials
  • Evaluate endpoint strategies (ORR, PFS, biomarker endpoints)
  • Determine regulatory pathways (breakthrough, orphan, accelerated approval)
  • Calculate sample sizes and enrollment timelines
  • Create safety monitoring plans
  • Compare trial design alternatives (single-arm vs. randomized)

Trigger phrases: "clinical trial design", "trial feasibility", "enrollment projections", "biomarker trial", "Phase 1/2 design", "basket trial", "endpoint selection", "regulatory pathway"

Quick Start

Basic Feasibility Assessment

from tooluniverse import ToolUniverse

tu = ToolUniverse(use_cache=True)
tu.load_tools()

# Example: Assess EGFR+ NSCLC trial feasibility
indication = "EGFR L858R+ non-small cell lung cancer"
biomarker = "EGFR L858R"

# 1. Disease prevalence
disease_info = tu.tools.OpenTargets_get_disease_id_description_by_name(
    diseaseName="non-small cell lung cancer"
)

# 2. Biomarker prevalence
variants = tu.tools.ClinVar_search_variants(
    gene="EGFR",
    significance="pathogenic"
)

# 3. Precedent trials
trials = tu.tools.search_clinical_trials(
    condition="EGFR positive non-small cell lung cancer",
    status="completed",
    phase="2"
)

# 4. Standard of care
soc_drug = tu.tools.drugbank_get_drug_basic_info_by_drug_name_or_id(
    drug_name_or_drugbank_id="osimertinib"
)

# Compile into feasibility report...

Files in This Skill

SKILL.md (Main Instructions)

  • Report-first approach - Create trial_feasibility_report.md FIRST
  • 6 research paths - Systematic data collection workflow
  • 14-section report structure - Executive summary to final recommendations
  • Evidence grading (A/B/C/D) - Grade all regulatory precedents
  • Feasibility scoring (0-100) - Quantitative assessment across 5 dimensions
  • Complete example workflow - Full EGFR+ NSCLC Phase 1/2 trial

EXAMPLES.md (5 Worked Examples)

  1. Biomarker-Selected Oncology Trial - EGFR L858R+ NSCLC (Score: 82/100, HIGH feasibility)
  2. Rare Disease Trial - Niemann-Pick Type C (Score: 58/100, MODERATE-LOW, slow enrollment)
  3. Superiority Trial vs. SOC - PD-1 inhibitor vs. pembrolizumab (Score: 87/100, HIGH)
  4. Non-Inferiority Trial - Oral anticoagulant (Score: 90/100, but large N, expensive)
  5. Basket Trial - NTRK fusion+ solid tumors (Score: 68/100, MODERATE, ultra-rare)

README.md (This File)

Quick start guide and overview

Key Features

Quantitative Feasibility Score (0-100)

Weighted composite across 5 dimensions:

  • Patient Availability (30%): Population size, biomarker prevalence, enrollment timeline
  • Endpoint Precedent (25%): FDA acceptance, measurement feasibility
  • Regulatory Clarity (20%): Pathway defined, precedent approvals
  • Comparator Feasibility (15%): SOC data, single-arm vs. randomized
  • Safety Monitoring (10%): Known toxicities, monitoring plan

Interpretation:

  • ≥75: HIGH feasibility - Recommend proceed
  • 50-74: MODERATE feasibility - Additional validation needed
  • <50: LOW feasibility - Significant de-risking required

Evidence Grading System

Grade Symbol Criteria Example
A ★★★ Regulatory acceptance, multiple precedents FDA-approved endpoint in indication
B ★★☆ Clinical validation, single precedent Phase 3 trial in related indication
C ★☆☆ Preclinical or exploratory Phase 1 use, biomarker validation
D ☆☆☆ Proposed, no validation Novel endpoint, no precedent

Report Structure (14 Sections)

  1. Executive Summary (with feasibility score)
  2. Disease Background
  3. Patient Population Analysis (with eligibility funnel)
  4. Biomarker Strategy
  5. Endpoint Selection & Justification
  6. Comparator Analysis
  7. Safety Endpoints & Monitoring Plan
  8. Study Design Recommendations
  9. Enrollment & Site Strategy
  10. Regulatory Pathway
  11. Budget & Resource Considerations
  12. Risk Assessment
  13. Success Criteria & Go/No-Go Decision
  14. Recommendations & Next Steps

Research Paths Overview

PATH 1: Patient Population Sizing

Objective: Calculate eligible patient pool and enrollment timeline

Tools:

  • OpenTargets_get_disease_id_description_by_name - Disease lookup
  • OpenTargets_get_diseases_phenotypes - Prevalence data
  • ClinVar_search_variants - Biomarker mutation frequency
  • gnomAD_search_gene_variants - Population genetics
  • PubMed_search_articles - Epidemiology literature

Outputs:

  • Annual eligible patients (with eligibility funnel)
  • Sites required
  • Enrollment timeline (months)

PATH 2: Biomarker Strategy

Objective: Assess biomarker testing feasibility and CDx landscape

Tools:

  • ClinVar_get_variant_details - Variant pathogenicity
  • COSMIC_search_mutations - Cancer mutation frequencies
  • PubMed_search_articles - CDx tests, testing guidelines

Outputs:

  • Biomarker prevalence (by geography, ethnicity)
  • Testing methods (NGS, IHC, liquid biopsy)
  • Turnaround time and cost

PATH 3: Comparator Selection

Objective: Identify standard of care and determine design (single-arm vs. randomized)

Tools:

  • drugbank_get_drug_basic_info_by_drug_name_or_id - Drug information
  • drugbank_get_indications_by_drug_name_or_drugbank_id - Approved indications
  • FDA_OrangeBook_search_drugs - Generic availability
  • search_clinical_trials - Historical control data

Outputs:

  • SOC drug(s) and efficacy
  • Single-arm vs. randomized recommendation
  • Comparator sourcing plan

PATH 4: Endpoint Selection

Objective: Select primary endpoint with regulatory precedent

Tools:

  • search_clinical_trials - Precedent trials, endpoints used
  • FDA_get_drug_approval_history - FDA acceptance by indication
  • PubMed_search_articles - Endpoint validation studies

Outputs:

  • Primary endpoint recommendation (ORR, PFS, DLT, biomarker)
  • Evidence grade (A/B/C/D)
  • Sample size calculation

PATH 5: Safety Monitoring

Objective: Design mechanism-based safety monitoring plan

Tools:

  • drugbank_get_pharmacology_by_drug_name_or_drugbank_id - Mechanism toxicity
  • FDA_get_warnings_and_cautions_by_drug_name - FDA warnings
  • FAERS_search_reports_by_drug_and_reaction - Real-world AEs
  • FAERS_count_reactions_by_drug_event - AE frequency

Outputs:

  • DLT definition (Phase 1)
  • Mechanism-based toxicities
  • Monitoring schedule (labs, imaging, ECG)
  • Stopping rules

PATH 6: Regulatory Pathway

Objective: Determine regulatory strategy and potential designations

Tools:

  • FDA_get_drug_approval_history - Precedent approvals
  • PubMed_search_articles - Breakthrough, orphan designations

Outputs:

  • Regulatory pathway (505(b)(1), 505(b)(2))
  • Designation opportunities (breakthrough, fast track, orphan)
  • Pre-IND meeting topics
  • IND timeline

Trial Design Types Supported

1. Biomarker-Selected Oncology

  • Example: EGFR L858R+ NSCLC
  • Design: Single-arm Phase 2, ORR primary
  • Feasibility: HIGH (clear biomarker, precedents)
  • Timeline: 12-18 months

2. Rare Disease

  • Example: Niemann-Pick Type C
  • Design: Single-arm vs. natural history
  • Feasibility: MODERATE-LOW (slow enrollment)
  • Timeline: 36-48 months
  • Special Considerations: Orphan drug, patient registries

3. Superiority Trial

  • Example: Novel PD-1 vs. pembrolizumab
  • Design: Randomized 1:1, ORR primary
  • Feasibility: HIGH (large population)
  • Timeline: 18-24 months

4. Non-Inferiority Trial

  • Example: Novel anticoagulant vs. apixaban
  • Design: Randomized, double-blind, event-driven
  • Feasibility: HIGH but expensive
  • Sample Size: Large (N=5,000+)

5. Basket Trial

  • Example: NTRK fusion+ solid tumors
  • Design: Single-arm, multiple histologies
  • Feasibility: MODERATE (ultra-rare, broad screening)
  • Timeline: 36-48 months

Common Workflows

Workflow 1: Quick Feasibility Check (1-2 hours)

  1. Disease prevalence (OpenTargets, PubMed)
  2. Biomarker frequency (ClinVar, gnomAD)
  3. Precedent trials (search_clinical_trials)
  4. Quick feasibility score
  5. Go/no-go recommendation

Use Case: Executive decision-making, portfolio prioritization

Workflow 2: Comprehensive Feasibility Report (1-2 days)

  1. Execute all 6 research paths
  2. Compile 14-section report
  3. Calculate enrollment funnel
  4. Regulatory pathway analysis
  5. Risk assessment
  6. Budget estimate

Use Case: Protocol development, investor presentations, FDA pre-IND prep

Workflow 3: Design Comparison (0.5-1 day)

  1. Assess 2-3 alternative designs (single-arm vs. randomized, different endpoints)
  2. Score each design
  3. Pros/cons analysis
  4. Recommendation

Use Case: Study team decision-making, choosing between design options

Performance Tips

  1. Enable caching: tu = ToolUniverse(use_cache=True) - Critical for repeated queries
  2. Parallel research paths: Run PATH 1-6 concurrently, not sequentially
  3. Use English terms: Always query tools in English, even if user asks in another language
  4. Cross-validate prevalence: Check ClinVar AND gnomAD AND literature for biomarkers
  5. Report-first: Create report structure FIRST, populate progressively
  6. Grade evidence: Every regulatory precedent needs evidence grade (A/B/C/D)

Prerequisites

Installation

pip install tooluniverse

Environment Variables

export OPENAI_API_KEY="sk-..."  # For LLM-based tool search
export NCBI_API_KEY="..."       # For higher PubMed rate limits (optional)

Recommended Knowledge

  • Clinical trial design basics (Phase 1/2/3, endpoints)
  • FDA regulatory pathways (IND, NDA, accelerated approval)
  • Biomarker concepts (CDx, NGS, prevalence)
  • Statistical concepts (sample size, power, non-inferiority margin)

Integration with Other Skills

Works well with:

  • tooluniverse-drug-research - Drug mechanism, preclinical data
  • tooluniverse-disease-research - Disease biology, natural history
  • tooluniverse-target-research - Target validation, druggability
  • tooluniverse-precision-oncology - Biomarker biology, resistance
  • tooluniverse-pharmacovigilance - Post-market safety data

Validation Checklist

Before recommending trial proceed:

Patient Population:

  • Prevalence data validated across ≥2 sources
  • Biomarker frequency confirmed (ClinVar, literature)
  • Eligibility criteria funnel calculated
  • Enrollment timeline realistic (<24 months for Phase 2)

Endpoints:

  • Primary endpoint has regulatory precedent (evidence grade A/B)
  • Measurement method standardized (RECIST, CTCAE, etc.)
  • Sample size calculation provided

Regulatory:

  • Pathway identified (505(b)(1), breakthrough, orphan)
  • Pre-IND meeting topics defined
  • Precedent approvals cited (drug names, years, NCT numbers)

Safety:

  • Mechanism-based toxicities identified
  • Monitoring schedule defined (labs, imaging frequency)
  • DLT definition provided (Phase 1)

Feasibility Score:

  • All 5 dimensions scored (patient, endpoint, regulatory, comparator, safety)
  • Rationale provided for each score
  • Overall score calculated (weighted average)

Success Metrics

HIGH Feasibility (≥75):

  • Patient availability strong (enrollment <18 months)
  • Endpoint has FDA precedent (grade A/B)
  • Clear regulatory path (precedents exist)
  • Comparator data robust (published trials)
  • Safety monitoring established (class effects known)

MODERATE Feasibility (50-74):

  • Patient availability moderate (enrollment 18-36 months)
  • Endpoint used in Phase 2 but not pivotal (grade B/C)
  • Regulatory path defined but needs FDA input
  • Comparator available but limited data
  • Safety monitoring feasible but novel mechanism

LOW Feasibility (<50):

  • Patient availability poor (enrollment >36 months or infeasible)
  • Endpoint novel, no precedent (grade D)
  • Regulatory path unclear
  • No comparator or historical data
  • Safety unknowns, high risk

Limitations

  1. Data Availability: Not all diseases/biomarkers have published prevalence data
  2. Geographic Variation: Prevalence estimates may vary by region (US vs. Asia)
  3. Enrollment Projections: Actual enrollment depends on site performance, competition
  4. Regulatory Landscape: FDA policies evolve; precedents are guidance, not guarantees
  5. Budget Estimates: Rough order-of-magnitude only; detailed budgets need finance input

Always: Validate feasibility findings with experienced clinical development team

Example Outputs

Example 1: EGFR+ NSCLC (HIGH Feasibility)

  • Score: 82/100
  • Recommendation: RECOMMEND PROCEED
  • Timeline: 24 months (enrollment + primary analysis)
  • Key Strengths: Large patient pool, ORR precedent, clear regulatory path

Example 2: Niemann-Pick Type C (MODERATE-LOW Feasibility)

  • Score: 58/100
  • Recommendation: CONDITIONAL GO (require registry partnership)
  • Timeline: 48-60 months
  • Key Challenge: Ultra-rare (36+ months enrollment)

Example 3: NTRK Basket (MODERATE Feasibility)

  • Score: 68/100
  • Recommendation: CONDITIONAL GO (require CGP partnership)
  • Timeline: 48 months (screening challenge)
  • Key Challenge: Ultra-rare biomarker (need broad NGS screening)

Citation

When using this skill, cite:

  • ToolUniverse: Gao S, Ding J, Zitnik M. ToolUniverse: Developing multi-tool AI systems with 750+ biomedical tools. arXiv:2024.xxxxx
  • Databases: OpenTargets, ClinVar, gnomAD, ClinicalTrials.gov, FDA, FAERS, DrugBank
  • Primary literature: Cite specific papers used for prevalence, endpoints

Support

Version

  • Version: 1.0.0
  • Last Updated: February 2026
  • Compatible with: ToolUniverse 0.5+
  • Focus: Early-phase trials (Phase 1/2 emphasis)

License

This skill follows ToolUniverse licensing. Check individual database terms of use for commercial clinical trial applications.

Quick Reference: Feasibility Score Calculation

# Feasibility Score Formula
dimensions = {
    'patient_availability': {'weight': 0.30, 'raw_score': 0-10},
    'endpoint_precedent': {'weight': 0.25, 'raw_score': 0-10},
    'regulatory_clarity': {'weight': 0.20, 'raw_score': 0-10},
    'comparator_feasibility': {'weight': 0.15, 'raw_score': 0-10},
    'safety_monitoring': {'weight': 0.10, 'raw_score': 0-10}
}

# Total feasibility score (0-100)
feasibility_score = sum(d['weight'] * d['raw_score'] * 10 for d in dimensions.values())

# Interpretation
if feasibility_score >= 75:
    recommendation = "RECOMMEND PROCEED"
elif feasibility_score >= 50:
    recommendation = "CONDITIONAL GO - Additional validation needed"
else:
    recommendation = "DO NOT RECOMMEND - Significant de-risking required"

Quick Reference: Evidence Grading

Grade Regulatory Clinical Preclinical Proposed
A ★★★ FDA-approved endpoint in indication Multiple Phase 3 precedents - -
B ★★☆ Used in Phase 3, not approved Single Phase 3 or multiple Phase 2 - -
C ★☆☆ Phase 1/2 only Case series Validated in animal models -
D ☆☆☆ No precedent Anecdotal Cell line data Computational only

Ready to assess your clinical trial feasibility? See SKILL.md for detailed instructions and EXAMPLES.md for 5 complete worked examples.