Sarcouncil Journal of Engineering and Computer Sciences

Abstract: This article explores cutting-edge advancements in generative AI (GenAI) for clinical decision support systems (CDSS). Unlike traditional rule-based CDSS, GenAI synthesizes multimodal data (EHRs, imaging, genomics) to generate dynamic care recommendations. The evolution from first-generation rule-based systems through machine learning approaches to current generative models represents a fundamental transformation in healthcare decision support. The technical architecture of GenAI systems incorporates large language models, multimodal integration frameworks, retrieval-augmented generation, FHIR-based data standardization, and explainable AI components. Real-world implementations demonstrate significant benefits, including Northwestern Memorial HealthCare's insurance summarization tool that reduced claim denials, an academic medical center's sepsis prediction system that improved early intervention rates, and Memorial Sloan Kettering's personalized oncology treatment planner. Emerging directions include federated learning for privacy-preserving collaboration, digital twins for treatment simulation, and ambient clinical intelligence systems that automate documentation while supporting clinical decision-making. Despite promising results, successful implementation requires addressing technical challenges while ensuring appropriate governance, clinical workflow integration, and stakeholder engagement.