Identifying the severity of diabetic retinopathy by visual function measures using both traditional statistical methods and interpretable machine learning: a cross-sectional study.

AIMS/HYPOTHESIS: To determine the extent to which diabetic retinopathy severity stage may be classified using machine learning (ML) and commonly used clinical measures of visual function together with age and sex. METHODS: We measured the visual function of 1901 eyes from 1032 participants in the Northern Ireland Sensory Ageing Study, deriving 12 variables from nine visual function tests. Missing values were imputed using chained equations. Participants were divided into four groups using clinical measures and grading of ophthalmic images: no diabetes mellitus (no DM), diabetes but no diabetic retinopathy (DM no DR), diabetic retinopathy without diabetic macular oedema (DR no DMO) and diabetic retinopathy with DMO (DR with DMO). Ensemble ML models were fitted to classify group membership for three tasks, distinguishing (A) the DM no DR group from the no DM group; (B) the DR no DMO group from the DM no DR group; and (C) the DR with DMO group from the DR no DMO group. More conventional multiple logistic regression models were also fitted for comparison. An interpretable ML technique was used to rank the contribution of visual function variables to predictions and to disentangle associations between diabetic eye disease and visual function from artefacts of the data collection process. RESULTS: The performance of the ensemble ML models was good across all three classification tasks, with accuracies of 0.92, 1.00 and 0.84, respectively, for tasks A-C, substantially exceeding the accuracies for logistic regression (0.84, 0.61 and 0.80, respectively). Reading index was highly ranked for tasks A and B, whereas near visual acuity and Moorfields chart acuity were important for task C. Microperimetry variables ranked highly for all three tasks, but this was partly due to a data artefact (a large proportion of missing values). CONCLUSIONS/INTERPRETATION: Ensemble ML models predicted status of diabetic eye disease with high accuracy using just age, sex and measures of visual function. Interpretable ML methods enabled us to identify profiles of visual function associated with different stages of diabetic eye disease, and to disentangle associations from artefacts of the data collection process. Together, these two techniques have great potential for developing prediction models using untidy real-world clinical data.

Assessment of the Classification of Age-Related Macular Degeneration Severity from the Northern Ireland Sensory Ageing Study Using a Measure of Dark Adaptation.

Purpose: To assess the differences in rod-mediated dark adaptation (RMDA) between different grades of age-related macular degeneration (AMD) severity using an OCT-based criterion compared with those of AMD severity using the Beckman color fundus photography (CFP)-based classification and to assess the association between the presence of subretinal drusenoid deposits (SDDs) and RMDA at different grades of AMD severity using an OCT-based classification. Design: Cross-sectional study. Participants: Participants from the Northern Ireland Sensory Ageing study (Queen's University Belfast). Methods: Complete RMDA (rod-intercept time [RIT]) data, CFP, and spectral-domain OCT images were extracted. Participants were stratified into 4 Beckman groups (omitting late-stage AMD) and 3 OCT-based groups. The presence and stage of SDDs were identified using OCT. Main Outcome Measures: Rod-intercept time data (age-corrected). Results: Data from 459 participants (median [interquartile range] age, 65 [59-71] years) were stratified by both the classifications. Subretinal drusenoid deposits were detected in 109 eyes. The median (interquartile range) RMDA for the Beckman classification (Beckman 0-3, with 3 being intermediate age-related macular degeneration [iAMD]) groups was 6.0 (4.5-8.7), 6.6 (4.7-10.5), 5.7 (4.4-7.4), and 13.2 (6-21.1) minutes, respectively. OCT classifications OCT0-OCT2 yielded different median (interquartile range) values: 5.8 (4.5-8.5), 8.4 (5.2-13.3), and 11.1 (5.3-20.1) minutes, respectively. After correcting for age, eyes in Beckman 3 (iAMD) had statistically significantly worse RMDA than eyes in the other Beckman groups (P ≤ 0.005 for all), with no statistically significant differences between the other Beckman groups. Similarly, after age correction, eyes in OCT2 had worse RMDA than eyes in OCT0 (P ≤ 0.001) and OCT1 (P < 0.01); however, there was no statistically significant difference between eyes in OCT0 and eyes in OCT1 (P = 0.195). The presence of SDDs was associated with worse RMDA in OCT2 (P < 0.01) but not in OCT1 (P = 0.285). Conclusions: Eyes with a structural definition of iAMD have delayed RMDA, regardless of whether a CFP- or OCT-based criterion is used. In this study, after correcting for age, the RMDA did not differ between groups of eyes defined to have early AMD or normal aging, regardless of the classification. The presence of SDDs has some effect on RMDA at different grades of AMD severity.