A systematic review published in *Radiology: Artificial Intelligence* indicates that while AI can generate economic value in radiology, evidence on its cost-effectiveness is limited and often unrealistic for real-world applications. Lead author Isabel Molwitz emphasized that the economic value of AI relies heavily on its implementation context. Most existing studies, which only a tiny fraction address economic outcomes, primarily focus on pixel-based tools rather than broader economic impacts. The commentary advocates for standardized frameworks and ongoing evaluations to better assess AI's financial outcomes against clinical needs. Co-author Eliot Siegel highlighted that significant hidden costs related to IT and training are frequently overlooked in existing economic models. The authors urge radiologists to prioritize AI tools that enhance workflow rather than merely focusing on image diagnosis, as this is where substantial economic benefits can lie, especially in critical applications like stroke detection. The review and commentary suggest that the radiology community should adopt a more realistic approach to AI's economic implications while recognizing the potential for improved cost-effectiveness through strategic implementations.
Tue, 26 May 2026 16:55:02 GMT | Radiological Society of North America | RSNA