TDCT – Trial Design Consistency Testing

Automated, reproducible protocol validation for detecting methodological, logical, and safety-related flaws in clinical trial protocols before participant enrolment.

TDCT was developed as part of the research paper:

Adeyemi S. (2025)
Automated Trial Design Consistency Testing (TDCT): A Validation System for Identifying Protocol Design Flaws Before Clinical Trial Initiation.

---

📌 Overview

TDCT (Trial Design Consistency Testing) is a modular, validator-driven framework that performs unit-test–style checks on clinical trial protocols.
It converts protocol content into structured YAML and automatically evaluates:

• Safety logic
• Eligibility rules
• Endpoint consistency
• Intervention design
• Measurement standardisation
• Regulatory/operational risks

This makes protocol review systematic, reproducible, and computational instead of manual-only.

---

⭐ Key Features

✔ Automated Protocol Scanning

Converts raw protocol text into structured YAML and runs >20 validators.

✔ Detects Critical Design Flaws

Identifies:

• safety gaps
• endpoint contradictions
• dosing errors
• missing exclusions
• measurement inconsistencies
• improper classifications

✔ Validated on Real Clinical Trials

TDCT was tested on:

Trial	Phase	N	Findings
HUNTER	Ib/IIa	60	7 findings
ACCORD	III	10,251	5 findings
SPRINT	III	9,361	7 findings

✔ Research Impact

Retrospective validation shows TDCT could have prevented or flagged:

• Early ACCORD termination (USD 45M impact)
• SPRINT BP measurement controversy
• Safety risks in the first-in-human HUNTER trial

---

📁 Repository Structure

tdct-validator/ │ ├── tdct/ │ ├── init.py │ ├── runner.py │ ├── utils.py │ ├── validators/ │ └── schema/ │ ├── protocols/ │ ├── hunter_protocol.yaml │ ├── accord_protocol.yaml │ └── sprint_protocol.yaml │ ├── findings/ │ ├── hunter_findings.json │ ├── accord_findings.json │ └── sprint_findings.json │ ├── examples/ │ ├── example_input.yaml │ └── example_output.json │ ├── CITATION.cff ├── requirements.txt ├── LICENSE └── README.md

yaml Copy code

---

⚙ Installation

git clone https://github.com/SamAde1203/tdct-validator.git
cd tdct-validator
pip install -r requirements.txt
▶ Running TDCT
Validate any protocol YAML:

bash
Copy code
python tdct/runner.py --protocol protocols/hunter_protocol.yaml
Findings will be saved to findings/ in JSON format.

🧩 Example YAML (Simplified)
yaml
Copy code
eligibility:
  age_min: 30
  age_max: 60
  exclusions:
    - severe_hypoglycemia_12m

intervention:
  doses:
    - 500mg
    - 1000mg
    - 1500mg

endpoints:
  primary: safety
  secondary:
    - HbA1c
    - fasting_glucose

measurement:
  BP_method: "not_specified"
🔍 Example Safety Validator
python
Copy code
class HypoglycemiaRiskValidator(BaseValidator):
    trigger = lambda p: p.population.age_max >= 65 and p.intervention.glucose_lowering

    def validate(self, protocol):
        exclusion = protocol.eligibility.exclusions
        if "severe_hypoglycemia_24m" not in exclusion:
            return self.critical(
                code="SAF-001",
                message="Missing severe hypoglycemia exclusion (24 months).",
                recommendation="Add exclusion: severe hypoglycemia within 24 months.",
                predicted_outcome="High risk of mortality in elderly on glucose-lowering therapy."
            )
📜 License
TDCT is released under the MIT License.
Commercial and academic use permitted with attribution.

📖 Citation
If you use TDCT in your research, please cite:

scss
Copy code
Adeyemi, S. (2025).
TDCT – Trial Design Consistency Testing (Version 1.0.0) [Computer software].
Zenodo. https://doi.org/10.5281/zenodo.17844210
🧾 Software Availability Statement (for your paper)
TDCT v1.0.0 is openly available under the MIT License at:

GitHub: https://github.com/SamAde1203/tdct-validator

Zenodo archive (DOI): https://doi.org/10.5281/zenodo.17844210

All protocol schemas, validators, and replication files for HUNTER, ACCORD, and SPRINT are included in the release.

🤝 Contributing
We welcome contributions including:

New validators

Auto-extraction NLP modules

Protocol schemas

TA-specific validator packs (oncology, CNS, paediatrics)

Submit a pull request or open an issue.

🙏 Acknowledgments
TDCT was created independently by Sam Adeyemi, MSc (UK).
No institutional funding was used.

Public protocol sources: ClinicalTrials.gov