Summary
Optimal CT stands for Optimization of Preoperative Treatment & Interactive Medical Assistance for Learning Cardiothoracic Surgery. The Optimal CT project is a collaboration with the Cardiothoracic Surgery Department at UK HealthCare. We are building a system that can intelligently communicate with patients during the weeks/days leading up to their scheduled surgery. This effort is in early stages.
The system can answer patient questions, provide health and wellness resources, and remind patients of important tasks such as stopping eating ahead of their procedure and coordinating transportation. The system’s primary function is to ensure medication adherence, particularly beta blockers, which are a crucial component of pre-op heart surgery care. The system asks patients to log their medication, gently reminding users if they haven’t responded.
The system is built using a Large Language Model (LLM) integrating Retrieval-Augmented Generation (RAG) techniques and Smart State protocols. These allow the system to manage complex interactions and automate communication. By providing patients with easy access to information and gentle reminders, the system aims to help patients adhere to pre-op protocols. For healthcare providers, the system strives to reduce administrative tasks.
OptimalCT is built on a Large Language Model architecture, accessed via the LLM Factory API. Its knowledge base is augmented using information from trusted sources like the Cardiothoracic Surgery Network (CTSNet) to provide accurate, context-aware answers.
Access
The following demo is just one piece of the Optimal CT project. The Demo can look up information pulled from the Cardiothoracic Surgery Network (CTSNet) website and share it with users in a conversational manner.
Ownership
The Optimal CT project is a collaboration with the Cardiothoracic Surgery Department at UK HealthCare. This effort is in early stages, with plans for deployment and publication.
Resources Utilized
Sam Armstrong has been the primary developer for OptimalCT. Developers Evan Damron and Mitchell Klusty have assisted on the project.
Resources include CAAI’s LLM Factory API.