Discrimination of indoor versus outdoor environmental state with machine learning algorithms in myopia observational studies.

BACKGROUND: Wearable smart watches provide large amount of real-time data on the environmental state of the users and are useful to determine risk factors for onset and progression of myopia. We aim to evaluate the efficacy of machine learning algorithm in differentiating indoor and outdoor locations as collected by use of smart watches. METHODS: Real time data on luminance, ultraviolet light levels and number of steps obtained with smart watches from dataset A: 12 adults from 8 scenes and manually recorded true locations. 70% of data was considered training set and support vector machine (SVM) algorithm generated using the variables to create a classification system. Data collected manually by the adults was the reference. The algorithm was used for predicting the location of the remaining 30% of dataset A. Accuracy was defined as the number of correct predictions divided by all. Similarly, data was corrected from dataset B: 172 children from 3 schools and 12 supervisors recorded true locations. Data collected by the supervisors was the reference. SVM model trained from dataset A was used to predict the location of dataset B for validation. Finally, we predicted the location of dataset B using the SVM model self-trained from dataset B. We repeated these three predictions with traditional univariate threshold segmentation method. RESULTS: In both datasets, SVM outperformed the univariate threshold segmentation method. In dataset A, the accuracy and AUC of SVM were 99.55% and 0.99 as compared to 95.11% and 0.95 with the univariate threshold segmentation (p < 0.01). In validation, the accuracy and AUC of SVM were 82.67% and 0.90 compared to 80.88% and 0.85 with the univariate threshold segmentation method (p < 0.01). In dataset B, the accuracy and AUC of SVM and AUC were 92.43% and 0.96 compared to 80.88% and 0.85 with the univariate threshold segmentation (p < 0.01). CONCLUSIONS: Machine learning algorithm allows for discrimination of outdoor versus indoor environments with high accuracy and provides an opportunity to study and determine the role of environmental risk factors in onset and progression of myopia. The accuracy of machine learning algorithm could be improved if the model is trained with the dataset itself.

Higher-order aberrations and their association with axial elongation in highly myopic children and adolescents.

BACKGROUND: Vision-dependent mechanisms play a role in myopia progression in childhood. Thus, we investigated the distribution of ocular and corneal higher-order aberrations (HOAs) in highly myopic Chinese children and adolescents and the relationship between HOA components and 1-year axial eye growth. METHODS: Baseline cycloplegic ocular and corneal HOAs, axial length (AL), spherical equivalent (SE), astigmatism and interpupillary distance (IPD) were determined for the right eyes of 458 highly myopic (SE ≤-5.0D) subjects. HOAs were compared among baseline age groups (≤12 years, 13-15 years and 16-18 years). Ninety-nine subjects completed the 1-year follow-up. Linear mixed model analyses were applied to determine the association between HOA components, other known confounding variables (age, gender, SE, astigmatism and IPD) and axial growth. A comparison with data from an early study of moderate myopia were conducted. RESULTS: Almost all ocular HOAs and few corneal HOAs exhibited significant differences between different age groups (all p<0.05). After 1 year, only ocular HOA components was significantly negative associated with a longer AL, including secondary horizontal comatic aberration (p=0.019), primary spherical aberration (p<0.001) and spherical HOA (p=0.026). Comparing with the moderate myopia data, the association of comatic aberration with AL growth was only found in high myopia. CONCLUSION: In highly myopic children and adolescents, lower levels of annual ocular secondary horizontal comatic aberration changes, besides spherical aberrations, were associated with axial elongation. This suggests that ocular HOA plays a potential role in refractive development in high myopia.

Design and Pilot data of the high myopia registration study: Shanghai Child and Adolescent Large-scale Eye Study (SCALE-HM).

PURPOSE: To describe the methodology and pilot data of the Shanghai Child and Adolescent Large-scale Eye Study (SCALE-HM). METHODS: This is a population-based, prospective, examiner-masked study with annual follow-up. Patients are 4- to 18-year-olds with high myopia. The participants will fill out questionnaires and then undergo visual acuity, axial length (AL), intraocular pressure, ophthalmologist assessment, microperimetry, cycloplegic refraction, Pentacam, wavefront aberration, fundus, blood and saliva examinations. To describe the pilot data, intergroup differences were assessed with t-tests or analysis of variance and a logistic regression model was used to determine the independent factors associated with peripapillary atrophy (PPA). RESULTS: Overall, 134 eyes of 79 participants met the pilot study recruitment criteria. The mean AL and spherical equivalent were 26.91 ± 1.07 mm and -9.40 ± 1.77 D, respectively. Peripapillary atrophy (PPA) (N = 112) and tessellated fundus (N = 67) were the most common fundus changes. The mean AL was significantly longer in PPA (27.08 ± 0.93 mm) than in non-PPA eyes (26.06 ± 1.31 mm; p < 0.001). Axial length (AL) (p = 0.041) was the only independent factor associated with PPA. Axial length (AL) was significantly longer in eyes with diffuse chorioretinal atrophy (N = 11; 28.02 ± 1.31 mm) than without myopic retinal lesions (N = 56; 26.48 ± 0.91 mm, p < 0.001) or with tessellated fundus (N = 67; 27.09 ± 0.97 mm, p = 0.012). The myopic degree was higher in eyes with diffuse chorioretinal atrophy than without myopic retinal lesions (-10.51 ± 2.76 D versus -9.06 ± 1.58 D, p = 0.039). CONCLUSION: Peripapillary atrophy and tessellated fundus were common in children and adolescents with high myopia. Results from this prospective study will help to understand the mechanisms, development and prognosis of these changes and can guide early myopia screening.

Crystalline Lens Power and Associated Factors in Highly Myopic Children and Adolescents Aged 4 to 19 Years.

PURPOSE: To investigate the distribution characteristics of lens power in highly myopic Chinese children and adolescents and its association with age, axial length (AL), and spherical equivalent (SE). DESIGN: Cross-sectional study. METHODS: A total of 459 highly myopic (SE ≤-5 diopter [D]) children and adolescents aged 4-19 years were included in the study. Participants underwent a series of ophthalmic examinations, which included AL, cycloplegic refraction, and Pentacam measurements. Lens power was calculated using Bennett's formula with its distribution described by age, AL, and SE. Multiple regression was conducted to analyze the associated factors of lens power. RESULTS: Greater lens power was independently associated with younger age, girls, shorter AL, and thicker lens thickness (standardized ß = -0.203, 0.214, -0.379 and 0.492, respectively; all P < .001). However, a significant difference in lens power with age was only found in participants younger than 9 years, after which it reached a plateau (mean difference of 1.23 and 0.084 D per age group, respectively). Lens power was negatively associated with AL only in participants with AL <27 mm. No correlation was observed between lens power and SE. CONCLUSION: Among highly myopic children and adolescents, differences in lens power with age declined significantly after 9 years of age, which was 1 year earlier than non-high myopic patients in previous studies, which implied differences in pathophysiological process between non-high myopia and high myopia. The decoupling of lens power and AL in eyes >27 mm might represent the limited influence of AL on lens power.

Prevalence of fundus tessellation and its associated factors in Chinese children and adolescents with high myopia.

PURPOSE: To investigate the prevalence and associated factors of fundus tessellation in highly myopic children and adolescents. METHODS: A total of 513 high myopes (spherical equivalent [SE] ≤ -5.0 D, 4-19 years of age) without any advanced pathological myopic lesions were enrolled. Fundus photographs and choroidal thickness (ChT) data were collected by SS-OCT. A novel grading approach was adopted to classify fundus tessellation into four categories on colour fundus photography, referring to the location of tessellation divided by an Early Treatment Diabetic Retinopathy Study grid centred on the fovea, through which closer to the fovea represents higher grades of fundus tessellation. Peripapillary atrophy (PPA) area and ovality index were also measured. RESULTS: Among the participants, with a mean age of 13.47 ± 3.13 years and mean SE of - 8.34 ± 1.91 D, there were 29 (5.7%), 95 (18.5%), 233 (45.4%) and 156 (30.4%) participants with grade 0 to grade 3 fundus tessellation, respectively. The ChT in both the macular and peripapillary area was negatively correlated with the fundus tessellation grade (R = -0.763 and -0.537, respectively, all p < 0.001). Higher grades of fundus tessellation were independently associated with thinner macular ChT (OR = 1.734, 95% CI: 1.621-1.856, p < 0.001), longer axial length (OR = 1.368, 95% CI: 1.105-1.695, p = 0.004), larger PPA area (OR = 1.391, 95% CI: 1.073-1.802, p = 0.013) and the female sex (OR = 1.605, 95% CI: 1.092-2.359, p = 0.016). CONCLUSION: The fundus tessellation grade could reflect the ChT, representing the severity of myopic maculopathy among young high myopes who rarely had any advanced lesions of pathological myopia. Fundus tessellation grade might be a potential index for assessing early-stage myopic maculopathy in children and adolescents.