Quality of life impact in patients with centre-involving diabetic macular oedema undergoing intravitreal injections using a novel AI-driven computerised adaptive testing system

Background Generic vision-related quality of life (VRQoL) instruments are commonly used to evaluate treatment outcomes in patients with diabetic macular oedema (DMO), but they may lack sensitivity to capture disease-specific impacts and treatment burden. This study aimed to evaluate changes in DMO-specific quality of life (QoL) following intravitreal therapy (IVT) for centre-involving DMO (CI-DMO) using RetCAT, a novel artificial intelligence-driven computerised adaptive testing system, and to compare its responsiveness with the National Eye Institute Visual Function Questionnaire (NEI VFQ-25). Methods This longitudinal real-world study was conducted at a tertiary eye centre in Milan, Italy. 100 treatment-naïve patients with CI-DMO were recruited between April 2023 and October 2024; 79 completed 12-month follow-up. Best-corrected visual acuity (BCVA), RetCAT (10 QoL domains) and NEI VFQ-25 (composite and 11 subscales) were assessed at baseline and 12 months. Vision change in the treated eye was categorised as ≥1 Early Treatment Diabetic Retinopathy Study (ETDRS) line gain (≥5 letters), no change or ≥1 ETDRS line loss (≥5 letters). Pre–post changes were analysed using paired tests and effect sizes. Results Following treatment, 24 patients reported a 1-line improvement in BCVA, which was associated with significant score improvements for 8 RetCAT domains (eg, MB 2.51±1.38 to 3.26±0.49, p=0.009; ES 0.76 and CN 1.09±0.54 to 1.65±0.54, p<0.001; ES 1.20). In comparison, only the NEI VFQ-25 composite score and its Near activities and Dependency domains showed significant improvements. Conclusions RetCAT detected broader and more sensitive quality of life improvements following IVT than the NEI VFQ-25, particularly in patients with modest visual acuity gains. It represents a valuable patient-centred outcome measure for monitoring treatment response in CI-DMO.