Avirup Guha, Director of Cardio-Oncology at Georgia Cancer Center, shared a post on X about a paper he co-authored with colleagues published in JACC: Advances:
“AI in Cardio-Oncology: A New Era
Large language models (LLMs) like GPT-4 can streamline cardio-oncology workflows – clinical note summarization, EHR data extraction, and trial matching – while maintaining accuracy and speed.
Modular AI Assistant Framework
Using retrieval-augmented generation (RAG), Azure Document Intelligence, LangChain and Streamlit, the proposed AI assistant integrates:
- Triage
- Cardiotoxicity monitoring
- Evidence synthesis
- Clinical decision support
All embedded in EHR systems.
Equity and Transparency First
AI must address:
- Bias mitigation across race, sex, and age
- Digital access disparities (43% low-income adults lack broadband)
- Federated learning for data privacy and inclusion in rural settings
- Ensuring equitable care in cardio-oncology.
Future Directions and Clinical Trials
LLMs like OncoLLM and TrialGPT already show:
- Accurate patient-trial matching
- Real-time eConsent chatbots
- Regulatory readiness (FDA lifecycle mgmt, HIPAA compliance)
AI – adaptive, inclusive, real-time cardio-oncology trials.”
Title: Transforming Cardio-Oncology Care Through AI-Driven Large Language Model Systems: A Roadmap for Future Implementation
Authors: Aryan Agar, Viraj Shah, Harikrishnan Hyma Kunhiraman, Tarek Nahle, Anant Madabhushi, Irbaz Bin Riaz, Momita Bandaru, Tochukwu M. Okwuosa, Nathanael Fillmore, Avirup Guha
You can read the Full Article in JACC: Advances.
More posts featuring Avirup Guha.
