Title: A User-Centered Approach to Customizing, Validating and Evaluating a Large Language Model for Empowering Healthcare Providers with Up-to-Date Information on Adolescent and Child Health
Background
Healthcare providers (HCPs) in Sub-Saharan Africa face challenges in accessing up-to-date information on adolescent and child health care which can hinder effective clinical practice. Large language models (LLMs) have the potential to address these challenges by providing up-to-date context-specific, evidence-based information.
Overall Objective
To co-design, validate and evaluate a context-specific, web-based large language model (LLM) tool aimed at enhancing the knowledge, skills, and practices of healthcare professionals (HCPs) regarding children and adolescent healthcare, ensuring regulatory compliance, and promoting health equity.
Specific Objectives
- Assess current knowledge, awareness, practices, and ethical concerns on the Utilization of Large Language Models in clinical practice.
- Iteratively customize, validate and evaluate a context-specific a web-based Large Language Model in clinical practice.
- Evaluate usability, user experience, and utility of a web-based Large Language Model in real-world clinical practice.
- Develop a proposed framework to guide regulatory compliance and quality assurance in Large Language model implementation in healthcare in Kenya.
Meditron
- A suite of open-source LLMs adapted for low-resource & humanitarian settings.
- Co-designed: with MDs from around the world and humanitarian actors at ICRC.
- Student-led: 50+ EPFL students for 1+ year
Large scale: 250 H100 GPUs for 1 week - Open: expert-curated, publicly available medical literature and global clinical practice guidelines.
Massive Online Open Validation and Evaluation (MOOVE)
The MOOVE creates granular evaluations that go beyond accuracy: including contextual appropriateness, empathy, alignment, and more.
- Log in to post comments
