Page 109 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Use Case 13: AI-Driven Optimization of Clinical Trial Preparation:

From Protocol Design to CRO Selection 4.1-Healthcare

Organization: MEDIAIPLUS Inc.
Country: Republic of Korea

Contact Person(s):
Primary contact: Nayeon Ryoo, biz@ mediaiplus .com
Secondary contact: Ji Hee Jung

1 Use Case Summary Table

Item Details

Category Healthcare
Problem Addressed Fragmented manual workflows in clinical trial design and Contract
Research Organization (CRO) selection lead to inefficiencies, high costs
and delays

Key Aspects of • Improve the speed and accuracy of clinical trial preparation
Solution • Optimize study design using LLM insights (MediC)
• Automate CRO matching with RAG (FiCRO)

Technology Large Language Models (LLM), Retrieval-Augmented Generation (RAG),
Keywords Natural Language Processing (NLP), Clinical Trial Optimization, AI Partner
Matching

Data Availability More than 720,000 structured and unstructured datasets from global clin-
ical trial registries, 29,500 CRO profiles, and biomedical research outputs

Metadata (Type of Structured tabular data on clinical trials and CROs, including protocols,
Data) phases, outcomes, regions and sponsor/CRO attributes.

Model Training and Custom LLMs fine-tuned on biomedical and clinical corpora; RAG models
Fine-Tuning enhanced using real-world trial matching data and continuous user feed-
back loops

Testbeds or Pilot Ongoing pilot deployments with biotech startups and clinical research
Deployments institutions in Korea; planned expansion with multinational CROs and
pharma sponsors

Code repositories Not publicly available due to data sensitivity and commercial confiden-
tiality.

73