Events

Mr. Minghong Yuan
Ph.D. Candidate in Management Information System
McCombs School of Business
The University of Texas at Austin

Artificial intelligence (AI) is increasingly used to augment physician decision-making, yet existing research has largely focused on algorithms that provide predictions or recommendations. We examine a distinct form of AI (i.e., information-generating AI) that produces novel diagnostic information while leaving decision authority with physicians. Using Medicare claims data from 2020 to 2024, we study physicians’ adoption of fractional flow reserve computed tomography (FFR-CT), an AI-enabled technology that derives functional assessments of coronary artery disease from coronary CT angiography (CCTA). We examine how the adoption and use of FFR-CT affect diagnostic decision quality and efficiency and how these effects evolve with physicians’ accumulated experience. We find that FFR-CT adoption improves diagnostic quality by reducing potentially avoidable invasive coronary angiography (ICA) procedures and enhances diagnostic efficiency by reducing subsequent functional testing. We further show that these improvements are partly explained by reductions in diagnostic uncertainty following adoption. Importantly, improvements in diagnostic decision quality become more pronounced as physicians accumulate experience with FFR-CT, whereas efficiency gains do not exhibit the same experience-dependent pattern. Our findings suggest that realizing the value of information-generating AI requires physicians to develop experience in using, interpreting and integrating AI-generated information into diagnosis decision-making.