From Algorithm to Bedside: Evaluating the Impact of Artificial Intelligence on Clinical Outcomes and Value-Based Care

As healthcare shifts to value-based care (VBC), managing populations has become increasingly complex, with Artificial Intelligence (AI) playing a key role. This scoping review evaluates real-world evidence of AI projects, focusing on their impact on healthcare outcomes, costs, and equity. Through a methodology involving searches of various databases, eligible studies comparing AI tools in diagnosis and workflow improvement with traditional care were analyzed. While the broader secondary literature suggests that certain predictive AI models hold theoretical potential to reduce mortality rates and shorten intensive care unit (ICU) durations, our scoping review of empirical applications reveals a different immediate reality. We found that the measurable, real-world value of AI is currently concentrated in operational and diagnostic proxy metrics such as workflow efficiency, diagnostic yield, and provider cognitive offloading rather than definitive systemic clinical outcomes. Furthermore, these empirical findings are often derived from highly heterogeneous pilot studies, limiting their broad generalizability. Real-world applications face systemic challenges, including alert fatigue, overdiagnosis risks, and biased predictive models. While AI holds potential for enhancing healthcare, significant gaps exist between theoretical advantages and actual clinical effectiveness. Future integration should emphasize robust implementation science, economic outcomes, and ethical frameworks to ensure health equity and create value beyond mere technical accuracy.