The power of predictive artificial intelligence for the detection of ectasia

Introduction Artificial intelligence (AI) is transforming the diagnosis and management of corneal ectatic diseases (CED) and ectasia risk assessment (ERA) before refractive surgery. Predictive models offer promise for earlier detection and personalized patient care.Areas covered This review summarizes AI applications in CED and ERA, covering keratoconus screening, staging, progression monitoring, and risk stratification. ERA is explained within the two-hit hypothesis, which suggests that intrinsic susceptibility (the first hit) and surgical impact (the second hit) are key factors. AI-based indices outperform traditional tools, particularly in detecting subclinical ectasia and identifying risk in fellow eyes after monocular surgery, and in very asymmetric ectasia. The TBIv2, created by the Brazilian BrAIN group, demonstrated the highest diagnostic accuracy. Among tomographic indices, BAD-Dv4 exceeded PRFI and BAD-Dv3. Population-optimized index (e.g. cTBI, cCBI) improved performance in Chinese cohorts. The Relational Tissue Altered (RTA) metric outperformed relational stromal bed and percent tissue altered (PTA) in assessing surgical risk.Expert opinion AI enhances the safety of refractive surgery through individualized risk assessment, but it should support, not replace, clinical judgment. Population-specific models are essential. Future advances include integrating layered tomography, biomechanics, genetic markers, and behavioral data, such as eye rubbing, to advance personalized ectasia management further.