Early-Phase Clinical Trial Protocol Generation using Generative AI
Life SciencesBusiness Impacts
Reduced protocol draft time from 3 weeks to 1 week
Ensured consistency in the initial draft
Decreased protocol revisions while enhancing quality
Customer Key Facts
- Country : USA
- Size : 10,000+
- Industry : Pharmaceutical Manufacturing - Lifesciences
- About : Customer is a global biopharmaceutical leader that focuses on discovering and delivering groundbreaking treatments for patients battling serious illnesses.
Elevating Early-Phase Clinical Trial Protocols to Precision with Generative AI
Clinical trial protocols, essential for drug trials, are currently drafted manually by CROs using client-provided templates and references. This initial drafting process typically takes about three weeks to complete. However, despite the extensive manual effort, these drafts often suffer from poor content quality, language inconsistencies, and various errors. Additionally, there is frequently a misalignment with client expectations, which leads to significant issues.
Challenges
- Developing the initial draft manually was a time-consuming process, taking three weeks.
- Drafting the content manually resulted in inconsistencies in language, spelling, and grammar, leading to poor content quality.
- Content misalignment alters meaning throughout the draft, causing confusion and necessitating more revisions.
Technologies Used
Google Cloud Storage
Vertex AI Embeddings for Text
PALM 2 (text-bision)
Gemini
Vertex AI workbench
Lang chain
Google docs API
Vector Search/Vector store
Solution
Quantiphi utilized LLM technology with engineered prompts to automate the creation of the initial draft for early clinical trial protocols. By leveraging sources such as the client's internal documents, historical protocols, and the investigator's brochure, the system generates tailored responses for each section of the study.
These responses are seamlessly integrated into the client's protocol model template, ensuring consistency and accuracy throughout the drafting process.
Results
- Reduced draft time from 3 weeks to 1 week: Leveraging generative AI streamlined the initial drafting process, enabling a faster turnaround time.
- Enhanced consistency: The AI-powered approach eliminated language inconsistencies and ensured adherence to the client's protocol template, simplifying review and approvals.
- Improved content quality: Automatic error detection and content alignment minimized spelling and grammatical errors, while also guaranteeing content aligns with client expectations.
- Breakthrough solution: The introduction of large language models (LLMs) has been a significant breakthrough in addressing challenges that have persisted for years. Quantiphi's expertise in LLMs, combined with our deep domain understanding, has made it possible to deliver a solution that effectively produces high-quality, consistent draft protocols.
