CUTTING-EDGE DIAGNOSTIC INNOVATIONS: A META-ANALYSIS OF EARLY KERATOCONUS DETECTION

MUJEEB UR REHMAN PARREY, HADEEL NAWAF ALENEZI, RASIL SALAH ALANAZI, MAHA M ABDUL-LATIF, MALIK AZHAR HUSSAIN, MOHAMMED M. ISMAIL, HAIDER OSMAN IBNIDRIS ELMISBAH, MUHAMMAD OMER AFZAL BHATTI

Background: Keratoconus (KC) is a progressive corneal ectasia that can remain undetected in its early or subclinical stages, delaying sight-preserving interventions such as corneal cross-linking (CXL). Accurate and timely diagnosis is essential to prevent progression and irreversible visual loss.

Objective: This systematic review and meta-analysis assessed the diagnostic accuracy of advanced technologies including corneal tomography, biomechanics, and Artificial Intelligence (AI)/Machine Learning (ML) for early or subclinical KC detection.

Methods: Following PRISMA-2020 and MOOSE guidelines, PubMed, Scopus, Web of Science, and Cochrane Library were searched (2015–2025) for English-language studies reporting diagnostic performance metrics for early KC using advanced modalities. Eligible studies were pooled using random-effects (DerSimonian–Laird) models to estimate sensitivity and specificity with 95% confidence intervals (CIs). Study quality was assessed using QUADAS-2 and the Newcastle–Ottawa Scale (NOS).

Results:Twenty-three studies encompassing 4,987 eyes met inclusion criteria. AI/ML-enhanced multimodal systems, particularly the Tomographic and Biomechanical Index (TBI), demonstrated superior diagnostic performance with pooled sensitivity of 0.94 (95% CI: 0.90–0.97) and specificity of 0.92 (95% CI: 0.88–0.95). These results exceeded the performance of single-modality tomography (sensitivity 0.86) and Placido-based topography (0.79). Subgroup analysis revealed the highest accuracy when combining Scheimpflug tomography and Corvis ST biomechanical parameters. Funnel plots indicated minimal publication bias.

Conclusion:AI-integrated multimodal diagnostic systems substantially enhance the early detection of keratoconus, marking a shift from morphology-based to data-driven precision diagnostics. Standardized validation across populations and devices is recommended to support clinical adoption.