Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 11 appraisals
PloS one
Fundus image analysis of retinitis pigmentosa using artificial intelligence
Retinitis pigmentosa (RP) is a group of inherited retinal diseases that are caused by genetic defects that lead to progressive photoreceptor loss and eventual blindness. Early diagnosis would be helpful for effective management of the disease; however, many patients stay unaware of early symptoms. Meanwhile, fundus images are widely obtained during routine medical checkups but are underused for detecting RP. This study explores the effectiveness of finetuning deep learning models, pre-trained for general visual tasks, to identify RP from color fundus images. The dataset comprised 321 color fundus images from 201 Japanese subjects at Keio University Hospital, including 200 images from 107 patients with retinitis pigmentosa and 121 images from 94 non-retinitis pigmentosa subjects. Multiple images were available for some subjects. Using transfer learning, pretrained convolutional neural network models -VGG16, Resnet50, and InceptionV3- were finetuned to detect RP. As a result, Inception V3 achieved the best accuracy of 96.97%, which matches the average diagnostic accuracy of ophthalmologists. Gradient-weighted Class Activation Mapping (Grad-CAM) suggested that the model attended to clinically relevant fundus regions, including the peripheral retina and posterior pole, which may reflect features such as peripheral degenerative changes and retinal vascular attenuation. These findings support the potential interpretability of the finetuned model and suggest that deep learning may assist ophthalmologists in RP screening as a supportive tool.
25 July 2026
Read appraisal →PloS one
AI-based color vision screening and educational counseling for high school students: An experimental study
Color vision deficiency (CVD) can negatively affect students' learning experiences and career choices; however, combined screening and educational counseling remain challenging in school settings, particularly in low-resource contexts. Conventional paper-based color vision tests are time-consuming, difficult to standardize, and rarely linked to individualized educational or career guidance. This study aimed to design, implement, and validate an Android-based mobile application that integrates CVD screening with artificial intelligence (AI)-assisted educational and career counseling for high school students. A cross-sectional study was conducted with 527 high school students in Danang, Viet Nam. The application incorporated a digital Ishihara-based screening module and an AI-assisted counseling component. Screening outcomes were compared with conventional Ishihara testing to evaluate agreement, while students' perceptions of usability, usefulness, and counseling relevance were assessed using a validated questionnaire. Statistical analyses included descriptive statistics, independent-samples t-tests, reliability analysis using Cronbach's alpha, and exploratory factor analysis. Seventeen students (3.23%) were identified as having CVD, including red-green, protan, deutan, and achromatopsia (complete CVD). The mobile application demonstrated high concordance with conventional screening results. No statistically significant differences were observed between students with and without CVD in overall system evaluation scores (p > 0.05). The questionnaire showed good internal consistency (Cronbach's alpha = 0.857) and a clear multi-factor structure reflecting students' acceptance of AI-assisted educational and career counseling. These findings indicate that the proposed mHealth application offers a feasible and scalable approach for school-based CVD screening combined with personalized educational and career guidance. The system may serve as a supportive decision-making tool for students and counselors, while further validation under diverse real-world conditions and across different mobile devices is recommended.
24 July 2026
Read appraisal →Medical image analysis
STAGE challenge: Structural-Functional Transition in Glaucoma Assessment
Glaucoma is a leading cause of irreversible yet preventable blindness in working-age populations. Clinical diagnos is currently relies on functional visual field (VF) examinations to evaluate visual function and monitor disease progression, but these tests are time-consuming and require close cooperation with ophthalmologists, limiting their practicality. Although deep learning has shown promise for glaucoma diagnosis, most models have focused on structural changes in fundus and OCT images without linking them to functional VF outcomes. To bridge this gap, the HDMI Laboratory, in collaboration with the Zhongshan Ophthalmic Center of Sun Yat-sen University, organized the STAGE Challenge: Structural-Functional Transition in Glaucoma Assessment. The challenge explores predicting functional VF indicators-mean deviation, sensitivity maps, and pattern deviation probability maps-directly from structural OCT images. A dataset of 401 OCT volumes (each with 256 cross-sectional images) is released with corresponding VF labels and demographic data, along with a standardized evaluation framework to ensure fair comparison. This paper summarizes the methods of the seven finalist teams and analyzes their results. All teams employed dual-branch architectures integrating OCT with tabular data, and those using task-specific OCT models achieved the highest performance. These findings highlight the importance of tailored deep learning strategies for linking structural imaging to functional outcomes. The STAGE Challenge thus establishes the first standardized benchmark with a large, curated dataset for this task, enabling systematic evaluation of algorithms for structure-function analysis in glaucoma, and providing a foundation for future research. Details of the competition are available at http://hdmilab.cn/competition/stage.
3 July 2026
Read appraisal →BMJ (Clinical research ed.)
Low concentration atropine eye drops and progression of myopia in children: multicentre placebo controlled, double masked, randomised trial in the UK (CHAMP-UK)
OBJECTIVES: To evaluate the efficacy and safety of low concentration atropine eye drops for reducing progression of myopia in children in the UK. DESIGN: Multicentre, double masked, superiority, placebo controlled, randomised trial. SETTING: National Health Service hospital eye services and academic institutions at five UK centres. PARTICIPANTS: 289 children aged 6-12 years with myopia between -0.50 and -10.0 dioptres (D). Participants were allocated in ratio of 2:1 to atropine or placebo. INTERVENTIONS: One eye drop of preserved atropine 0.01% or placebo daily for two years. MAIN OUTCOME MEASURES: The primary outcome was spherical equivalent refractive error of both eyes measured by autorefractor under cycloplegia after two years. Secondary outcomes included change in axial length, best corrected distance and near visual acuity, reading speed, pupil diameter, spectacle correction, adverse event rates, quality of life, and tolerability. Outcomes were collected every six months. An electronic monitoring system was used to assess adherence. RESULTS: 192 participants were included in the atropine group and 97 in the placebo group, with an average age of 9.3 years (standard deviation (SD) 1.7 years). 207 (72%) reported white ethnicity, 161 (56%) were girls, and the mean level of myopia was -2.87 D (SD 1.71 D). A total of 235 (81%) participants completed the study, with primary outcome data available for 230 (80%) participants: 151 (79%) in the atropine group and 79 (81%) in the placebo group. Atropine eye drops were more effective than placebo in reducing myopia progression (mean difference 0.33 D, 95% confidence interval (CI) 0.17 to 0.49 D, P<0.001). Prespecified subgroup analyses did not show differences according to age, ethnicity, sex, or severity of myopia. Changes in central axial length were significantly less in the atropine group versus placebo group: mean difference 0.14 mm (95% CI 0.07 to 0.21, P<0.001). There were no differences in other secondary outcomes, except pupil diameter, which was greater in the atropine group (0.36 mm, 95% CI 0.17, 0.55, P<0.001), and no differences in frequency of adverse events or in tolerability measures. No serious adverse events were related to the trial drugs. CONCLUSIONS: Low concentration atropine (0.01%) eye drops significantly reduced progression of myopia and were well tolerated compared with placebo in children in the UK. TRIAL REGISTRATION: ISRCTN registry ISRCTN99883695, ClinicalTrials.gov NCT03690089.
13 June 2026
Read appraisal →Experimental eye research
Automated segmentation of pterygium lesions using multiscale deep learning networks
Pterygium is an eye condition that needs to be identified at an early stage so that its progression can be mitigated in order to avoid the possible threat of visual impairment. One of the important low-level modules in determining the severity level of the pterygium automatically is to measure the extent of fibrovascular tissue encroachment onto the corneal and pupil regions. Therefore, it is important to develop a semantic segmentation method to map and quantify the lesion tissues accurately, especially through the deep learning technique. As the pterygium condition progresses from the trace stage to the severe stage, the lesions will generally become bigger but maintain more or less the same shape, which looks like a wedge. Taking inspiration from this unique pattern of pterygium lesions, this work aims to explore multiscale deep learning networks to capture the variable scales of the lesion features effectively. All the proposed multiscale modules are embedded at the bottleneck layer of the base UNet architecture, whereby the feature map size is the smallest to reduce the potential increment in the computational burden. Apart from that, the total number of convolutional channels is fixed to 1024, whereby the multiscale module replaces the first convolutional set at the UNet's bottleneck layer. Two main multiscale modules have been explored, which are spatial pyramid pooling (SPP) and atrous spatial pyramid pooling (ASPP) modules. Furthermore, each of these modules has been expanded by analyzing the equal-flow (EF) and waterfall-flow (WF) patterns in constructing the parallel paths that represent the multiscale features. According to the simulation results, the best segmentation network is UNet, which has been embedded with three parallel paths of the EF-ASPP module that produces the lowest Hausdorff distance of 16.75 pixels. As a result, the severity level of pterygium can be better predicted through accurate extraction of the lesion map. This work can be further improved by analyzing different network architectures of the multiscale module, especially networks that focuses on the larger feature maps size, but surely with a trade-off in computational burden.
5 May 2026
Read appraisal →PloS one
Machine learning identifies pupil size and corneal thickness as key predictors of axial elongation rate
PURPOSE: This study aimed to develop a machine learning-based prediction model for myopia progression using ocular biometric parameters to provide an objective assessment tool for clinical practice. METHODS: A retrospective analysis was conducted on patients treated at Shanghai Parkway Health Ophthalmology Department as the training set, and myopic individuals from the Optometry Center of Peking University People's Hospital as the validation set. Demographic and biometric data were collected, including central corneal thickness (CCT), axial length (AL), corneal curvature (K-value), anterior chamber depth (ACD), corneal diameter (WTW), and pupil size (PS). Seven machine learning models (e.g., XGBoost, random forest, support vector machine) were employed for modeling, with performance optimized via 5-fold cross-validation. Model accuracy was evaluated using mean squared error (MSE) and the coefficient of determination (R²), and variable importance was analyzed. RESULTS: No statistically significant differences were observed in baseline characteristics between the training and validation sets (all P > 0.05). The XGBoost model demonstrated the best performance, achieving R² = 0.913 (MSE = 0.005) on the training set and R² = 0.766 (MSE = 0.016) on the test set. Variable importance analysis revealed pupil size (score 100) and corneal thickness (40.88) as the key predictors of axial elongation rate, followed by age of onset (17.96). CONCLUSION: The machine learning-based prediction model effectively utilizes ocular biometric data to assess myopia progression risk, with pupil size and corneal thickness identified as core predictive factors. This model provides a quantitative tool for early clinical intervention. Future studies should expand the sample size and incorporate additional biomarkers to optimize performance.
27 Apr 2026
Read appraisal →The British journal of ophthalmology
Ethnic and dose-dependent differences in atropine efficacy for myopia control: a systematic review and meta-analysis
BACKGROUND/AIM: Clinical uncertainty remains regarding optimal atropine concentration, treatment duration and potential differences in efficacy for myopia control between Asian and non-Asian children. This systematic review and meta-analysis evaluated the efficacy of different concentrations of atropine for myopia control, comparing outcomes among East Asian, South Asian and white European children. METHODS: Five databases were searched for randomised controlled trials (RCTs) including children ≤16 years with myopia who received atropine treatment. Thirty-four RCTs with ≥12 months of follow-up were included. Weighted mean differences (WMD) in spherical equivalent refraction (SER) progression and axial length (AL) elongation were pooled by atropine concentration and ethnicity. RESULTS: Compared with controls, atropine significantly reduced myopia progression across all concentrations: <0.1% (WMD in SER: 0.44 (95% CI 0.35 to 0.52) dioptres (D)/year; AL: -0.20 (95% CI -0.24 to -0.16) mm/year), 0.1% to <0.5% (0.81 (95% CI 0.50 to 1.13) D/year) and ≥0.5% (1.06 (95% CI 0.88 to 1.24) D/year; -0.36 (95% CI -0.40 to -0.33) mm/year). The pooled effect on SER and AL progression across all concentrations was greater in East Asians (0.63 (95% CI 0.50 to 0.76) D/year; -0.26 (95% CI -0.31 to -0.20) mm/year) than in South Asians (0.40 (95% CI 0.11 to 0.70) D/year; -0.13 (95% CI -0.21 to -0.05) mm/year) or white Europeans (0.18 (95% CI 0.11 to 0.25) D/year; -0.11 (95% CI -0.16 to -0.05) mm/year). CONCLUSION: Atropine slows myopia progression in a dose-dependent manner in studies with 1-5 years of follow-up. Efficacy appears greater in Asian children, particularly East Asians, who also exhibit greater photopic pupil dilation. These findings support the role of atropine in myopia control and highlight the importance of ethnicity-specific considerations when prescribing and tailoring treatment strategies. PROSPERO REGISTRATION NUMBER: CRD42023454104.
26 Apr 2026
Read appraisal →The British journal of ophthalmology
Two-year outcomes of repeated red light therapy in premyopic children: sustained efficacy and rebound effects
BACKGROUND: To evaluate the long-term prevention effectiveness and rebound effect of repeated red light therapy (RRLT) in children with premyopia over 2 years. METHODS: A total of 108 premyopic children (cycloplegia spherical equivalent refraction (SER): -0.50 to +0.75 D) were enrolled and followed for 24 months. Participants were randomly assigned to the RRLT or control groups. The RRLT was administered two times per day for 3 min per session, with at least 4-hour interval. At the beginning of the second year, participants receiving RRLT were further randomised into continued treatment and washout subgroups. Axial length (AL), SER and subfoveal choroidal thickness (SChT) were measured. RESULTS: Over 2 years, the RRLT group showed significantly smaller AL elongation (0.26 mm; 95% CI 0.18 to 0.35 mm) and SER progression (-0.21 D; 95% CI -0.35 to -0.08 D) compared with the controls (AL: 0.43 mm; 95% CI 0.36 to 0.49 mm; SER: -0.66 D; 95% CI -0.79 to -0.52 D). The RRLT group also demonstrated significantly less SChT thinning (-2.44 µm; 95% CI -16.11 to 11.23 µm) than the controls (-44.12 µm; 95% CI -53.05 to -35.19 µm). After RRLT cessation in the second year, the washout subgroup exhibited significantly faster AL elongation and more SChT thinning than the controls, with no significant difference in SER progression. CONCLUSIONS: The 2-year RRLT intervention effectively retarded AL elongation and SER progression in premyopic children by 0.17 mm and -0.45 D, respectively. Notably, a significant rebound effect was observed in AL growth following 1-year RRLT cessation.
25 Apr 2026
Read appraisal →PloS one
Safety and efficacy of Vectored thermal pulsation, Intense pulsed light, and Eyelid Warm compress therapies (VIEW) for meibomian gland dysfunction: Study design and baseline characteristics of a randomized controlled trial
BACKGROUND: Meibomian gland dysfunction (MGD) is one of the most prevalent chronic eye diseases, characterized by meibomian gland obstruction and abnormal meibum secretion both in quantity and quality. OBJECTIVE: To assess the efficacy and safety of vectored thermal pulsation (VTP), intense pulsed light plus meibomian gland expression (IPL + MGX), and eyelid warm compress (EW). METHODS: This is a 3-arm, assessor-masked, randomized controlled trial. 360 symptomatic mild to moderate MGD subjects would be recruited and randomized into 3 intervention groups. Subjects from Group-A, Group-B, and Group-C would be treated with one-session VTP (at baseline), four-session IPL + MGX (at baseline, month-1, month-2, and month-3 visits), and twice-daily EW for 15 months, respectively. All subjects would be prescribed topical artificial tears (Hypromellose, 0.3% w/v) to be used as frequently as needed from wash-out (1 month before baseline) to endpoint (month-15 visit). Subjects would be assessed 9 times throughout a total of 16-month study period. Five self-administered questionnaires would be collected each time in random order. Masked observers would conduct examinations using slit-lamp biomicroscopy, LipiView® II, and Keratograph 5M machines. The eye with worse MGD would be selected and analyzed as the study eye. Tear film break-up time (TFBUT) at month-6 and month-15 visits are the primary outcomes. RESULTS: 374 subjects were enrolled and randomized into three groups, and 372 subjects (aged 54.4 ± 14.7 years) were finally analyzed. The mean of TFBUT time at baseline was 3.1 ± 1.7 seconds. CONCLUSION: This is the first head-to-head comparison of three physical treatments for mild to moderate MGD on their therapeutic efficacy, safety, longevity and compliance in a randomized controlled trial setting.
5 Apr 2026
Read appraisal →Journal of global health
Exposure-response association between outdoor activity time and myopia risk in Chinese children and adolescents: a systematic review and meta-analysis
BACKGROUND: The prevalence of myopia among Chinese children and adolescents has increased sharply over recent decades, raising concern about modifiable environmental risk factors. Although outdoor activity is widely recognised as protective, the quantitative exposure-response relationship within the Chinese educational context remains incompletely defined. METHODS: A systematic review and meta-analysis were conducted following PRISMA guidelines. Studies examining the association between daily outdoor activity time and myopia risk among Chinese youth were identified from major English and Chinese databases (Web of Science, PubMed, CNKI, and Wanfang) through October 2025. Random-effects meta-analysis was conducted to pool odds ratios (ORs). Prespecified subgroup and sensitivity analyses examined exposure thresholds, study design, and geographic latitude. RESULTS: We included 31 studies comprising 380 215 participants. Greater outdoor activity was consistently associated with lower myopia risk (pooled OR = 0.75; 95% confidence interval (CI) = 0.71-0.80), although substantial heterogeneity was observed (I2 = 94%). Compared with <1 hour per day, outdoor exposure of 1-2 hours (OR = 0.85; 95% CI = 0.79-0.92), 2-3 hours (OR = 0.86; 95% CI = 0.78-0.95), and > 3 hours (OR = 0.74; 95% CI = 0.63-0.87) was associated with reduced risk. When dichotomised at two hours, exposure >2 hours yielded an OR of 0.74 (95% CI = 0.69-0.80). Findings were directionally consistent across study designs and latitude strata. CONCLUSIONS: Greater outdoor activity time is associated with lower myopia risk among Chinese children and adolescents. An approximate daily exposure of around two hours appears to represent a pragmatic benchmark under current educational conditions.
4 Apr 2026
Read appraisal →Clinical & experimental optometry
Advancements in artificial intelligence for meibography: clinical utility and contributions to meibomian gland dysfunction management
Recent advances in artificial intelligence (AI) have enhanced the capabilities of meibography by enabling objective and quantitative assessment of meibomian gland structure. This review explores the clinical utility of AI-based meibography in the diagnosis and management of Meibomian Gland Dysfunction (MGD), with a focus on segmentation, morphological analysis, and disease staging. Developments in deep learning have enabled more precise gland feature extraction, including gland dropout, density, and tortuosity, supporting efforts towards standardised and reproducible clinical evaluation. Although not the focus of this review, insights from traditional image processing techniques are referenced to highlight potential areas of improvement in current AI models. Key issues such as limited modelling of regional gland variation, restricted dataset diversity, and lack of standardised image quality control are discussed. Although significant progress has been made, further work is needed to ensure AI-driven meibography tools are generalisable, interpretable, and suitable for broad clinical implementation.
4 Apr 2026
Read appraisal →