Abstract
The treatment of diabetic macular oedema (DMO) likely involves a series of treatment episodes separated by gaps during which no vascular endothelial growth factor inhibitor injections are delivered. Our aim is to differentiate the episodes and gaps with the isolation forests algorithm using the Fight Retinal Blindness! registry. We analysed 11 786 injection intervals (12 803 injections) and found that the period between adjacent injections≥38 weeks (95% CI 34 to 43 weeks) apart could be regarded as a treatment gap. The results will allow treatment episodes to be isolated so that the treatment patterns of DMO can be characterised more accurately.