Platform: AI powered application to review Clinical Trail Applications
Sector: Clinical Research, HealthTech, Ethics Governance
Technology Stack: HuggingFace Transformers, LangChain, Tesseract OCR, BERT, T5, LLaMA, Flan-T5, Python
Compliance: GDPR, POPIA, GCP
Project Overview
Zepto Digital Labs designed and developed — an AI-powered ethics review engine integrated into the Clinical Trail Applications platform. The solution helps Research Ethics Committees (RECs) in low- and middle-income countries review clinical research applications more efficiently and consistently.
Problem
RECs were overwhelmed with:
Increasing volume and complexity of submissions
Manual document navigation and data validation
Inconsistent reviews and slow approval timelines (2–6 months)
High administrative costs and cognitive load
Our Solution
- We built Review Ethics Artificial Intelligence — a secure, private-cloud AI engine that supports RECs by:
- Extracting & summarizing lengthy documents (protocols, consent forms)
- Validating consistency between applicant inputs and attached files
- Drafting reviewer responses using AI (editable & explainable)
- Alerting reviewers about missing or conflicting information
- Automating repetitive tasks without replacing human judgment
Technical Highlights
OCR: Tesseract used for converting scanned submissions into searchable content
NLP Summarization: BART & T5 for document summaries
Q&A Validation: BERT, DistilBERT to check answer-document consistency
Response Generation: LLaMA & Flan-T5 for drafting reviewer form inputs
Workflow Orchestration: LangChain for chaining AI tasks securely
On-Prem AI Deployment: All models trained and deployed within secure, client-owned infrastructure to ensure zero data leakage
Results
60–70% reduction in reviewer time per application
>85% accuracy in automated answer validation
Improved consistency and reviewer satisfaction
Faster decisions, lower costs for RECs, and higher confidence in data integrity
Strategic Impact
Helps standardize reviews across RECs in Africa and LMICs
Builds local capacity to handle complex submissions
Sets a new benchmark in secure, ethical AI adoption in health research
Looking for AI Solutions for your business?
Whether you're in healthtech, compliance, EdTech, or legal review automation, we can help build secure, domain-specific AI tools that augment human experts and ensure high standards of transparency, accuracy, and compliance.
Let’s connect if you need:
NLP & document summarization
On-prem AI deployment
Workflow automation using AI
Privacy-compliant AI tools