This research introduces a novel approach to adapt powerful multimodal models for deployment in computationally-limited environments like healthcare facilities.

• Combines foundational models with supervised ML to enable automated diagnosis and treatment planning
• Uses contrastive learning techniques to efficiently adapt multimodal embeddings to specific domains
• Addresses the critical challenge of limited onsite computational resources in healthcare settings
• Demonstrates how complex AI systems can be optimized for real-world medical applications

This research is particularly valuable for medical institutions seeking to implement advanced AI capabilities without requiring massive computational infrastructure, potentially democratizing access to cutting-edge diagnostic tools.

Efficient Domain Adaptation of Multimodal Embeddings using Constrastive Learning