Benchmarking spatial clustering methods with spatially resolved transcriptomics data.

Spatial clustering, which shares an analogy with single-cell clustering, has expanded the scope of tissue physiology studies from cell-centroid to structure-centroid with spatially resolved transcriptomics (SRT) data. Computational methods have undergone remarkable development in recent years, but a comprehensive benchmark study is still lacking. Here we present a benchmark study of 13 computational methods on 34 SRT data (7 datasets). The performance was evaluated on the basis of accuracy, spatial continuity, marker genes detection, scalability, and robustness. We found existing methods were complementary in terms of their performance and functionality, and we provide guidance for selecting appropriate methods for given scenarios. On testing additional 22 challenging datasets, we identified challenges in identifying noncontinuous spatial domains and limitations of existing methods, highlighting their inadequacies in handling recent large-scale tasks. Furthermore, with 145 simulated data, we examined the robustness of these methods against four different factors, and assessed the impact of pre- and postprocessing approaches. Our study offers a comprehensive evaluation of existing spatial clustering methods with SRT data, paving the way for future advancements in this rapidly evolving field.

Unique composition of ocular surface microbiome in the old patients with dry eye and diabetes mellitus in a community from Shanghai, China.

BACKGROUND: This study investigates the variations in microbiome abundance and diversity on the ocular surfaces of diabetic patients suffering from dry eye within a community setting. The goal is to offer theoretical insights for the community-level prevention and treatment of dry eye in diabetic cohorts. METHODS: Dry eye screening was performed in the Shanghai Cohort Study of Diabetic Eye Disease (SCODE) from July 15, 2021, to August 15, 2021, in the Xingjing community; this study included both a population with diabetes and a normal population. The population with diabetes included a dry eye group (DM-DE, n = 40) and a non-dry eye group (DM-NoDE, n = 39). The normal population included a dry eye group (NoDM-DE, n = 40) and a control group (control, n = 39). High-throughput sequencing of the 16 S rRNA V3-V4 region was performed on conjunctival swab from both eyes of each subject, and the composition of microbiome on the ocular surface of each group was analyzed. RESULTS: Significant statistical differences were observed in both α and ß diversity of the ocular surface microbiome among the diabetic dry eye, diabetic non-dry eye, non-diabetic dry eye, and normal control groups (P < 0.05). CONCLUSIONS: The study revealed distinct microecological compositions on the ocular surfaces between the diabetic dry eye group and other studied groups. Firmicutes and Anoxybacillus were unique bacterial phyla and genera in the dry eye with DM group, while Actinobacteria and Corynebacterium were unique bacterial phyla and genera in the normal control group.

Evaluating imaging repeatability of fully self-service fundus photography within a community-based eye disease screening setting.

PURPOSE: This study aimed to investigate the imaging repeatability of self-service fundus photography compared to traditional fundus photography performed by experienced operators. DESIGN: Prospective cross-sectional study. METHODS: In a community-based eye diseases screening site, we recruited 65 eyes (65 participants) from the resident population of Shanghai, China. All participants were devoid of cataract or any other conditions that could potentially compromise the quality of fundus imaging. Participants were categorized into fully self-service fundus photography or traditional fundus photography group. Image quantitative analysis software was used to extract clinically relevant indicators from the fundus images. Finally, a statistical analysis was performed to depict the imaging repeatability of fully self-service fundus photography. RESULTS: There was no statistical difference in the absolute differences, or the extents of variation of the indicators between the two groups. The extents of variation of all the measurement indicators, with the exception of the optic cup area, were below 10% in both groups. The Bland-Altman plots and multivariate analysis results were consistent with results mentioned above. CONCLUSIONS: The image repeatability of fully self-service fundus photography is comparable to that of traditional fundus photography performed by professionals, demonstrating promise in large-scale eye disease screening programs.

Remnant cholesterol is correlated with retinal vascular morphology and diabetic retinopathy in type 2 diabetes mellitus: a cross-sectional study.

BACKGROUND: The association between remnant cholesterol (RC) and diabetic retinopathy (DR) in type 2 diabetes mellitus (T2DM) remains unclear. Morphological changes in retinal vessels have been reported to predict vascular complications of diabetes, including DR. METHODS: This cross-sectional study included 6535 individuals with T2DM. The RC value was calculated using the recognized formula. The retinal vascular parameters were measured using fundus photography. The independent relationship between RC and DR was analyzed using binary logistic regression models. Multiple linear regression and subgroup analyses were employed to investigate the link between RC and vascular parameters, including the retinal arteriolar diameter (CRAE), venular diameter (CRVE), and fractal dimension (Df). Mediation analysis was performed to assess whether the vascular morphology could explain the association between RC and DR. RESULTS: RC was independently associated with DR in patients with a longer duration of T2DM (> 7 years). Patients with the highest quartile RC levels had larger CRAE (5.559 [4.093, 7.025] µm), CRVE (7.620 [5.298, 9.941] µm) and Df (0.013 [0.009, 0.017]) compared with patients with the lowest quartile RC levels. Results were robust across different subgroups. The association between RC and DR was mediated by CRVE (0.020 ± 0.005; 95% confidence interval: 0.012-0.032). CONCLUSIONS: RC may be a risk factor for DR among those who have had T2DM for a longer period of time. Higher RC levels were correlated with wider retinal arterioles and venules as well as higher Df, and it may contribute to DR through the dilation of retinal venules.

Service Quality and Residents' Preferences for Facilitated Self-Service Fundus Disease Screening: Cross-Sectional Study.

BACKGROUND: Fundus photography is the most important examination in eye disease screening. A facilitated self-service eye screening pattern based on the fully automatic fundus camera was developed in 2022 in Shanghai, China; it may help solve the problem of insufficient human resources in primary health care institutions. However, the service quality and residents' preference for this new pattern are unclear. OBJECTIVE: This study aimed to compare the service quality and residents' preferences between facilitated self-service eye screening and traditional manual screening and to explore the relationships between the screening service's quality and residents' preferences. METHODS: We conducted a cross-sectional study in Shanghai, China. Residents who underwent facilitated self-service fundus disease screening at one of the screening sites were assigned to the exposure group; those who were screened with a traditional fundus camera operated by an optometrist at an adjacent site comprised the control group. The primary outcome was the screening service quality, including effectiveness (image quality and screening efficiency), physiological discomfort, safety, convenience, and trustworthiness. The secondary outcome was the participants' preferences. Differences in service quality and the participants' preferences between the 2 groups were compared using chi-square tests separately. Subgroup analyses for exploring the relationships between the screening service's quality and residents' preference were conducted using generalized logit models. RESULTS: A total of 358 residents enrolled; among them, 176 (49.16%) were included in the exposure group and the remaining 182 (50.84%) in the control group. Residents' basic characteristics were balanced between the 2 groups. There was no significant difference in service quality between the 2 groups (image quality pass rate: P=.79; average screening time: P=.57; no physiological discomfort rate: P=.92; safety rate: P=.78; convenience rate: P=.95; trustworthiness rate: P=.20). However, the proportion of participants who were willing to use the same technology for their next screening was significantly lower in the exposure group than in the control group (P<.001). Subgroup analyses suggest that distrust in the facilitated self-service eye screening might increase the probability of refusal to undergo screening (P=.02). CONCLUSIONS: This study confirms that the facilitated self-service fundus disease screening pattern could achieve good service quality. However, it was difficult to reverse residents' preferences for manual screening in a short period, especially when the original manual service was already excellent. Therefore, the digital transformation of health care must be cautious. We suggest that attention be paid to the residents' individual needs. More efficient man-machine collaboration and personalized health management solutions based on large language models are both needed.

Medical Staff and Resident Preferences for Using Deep Learning in Eye Disease Screening: Discrete Choice Experiment.

BACKGROUND: Deep learning-assisted eye disease diagnosis technology is increasingly applied in eye disease screening. However, no research has suggested the prerequisites for health care service providers and residents willing to use it. OBJECTIVE: The aim of this paper is to reveal the preferences of health care service providers and residents for using artificial intelligence (AI) in community-based eye disease screening, particularly their preference for accuracy. METHODS: Discrete choice experiments for health care providers and residents were conducted in Shanghai, China. In total, 34 medical institutions with adequate AI-assisted screening experience participated. A total of 39 medical staff and 318 residents were asked to answer the questionnaire and make a trade-off among alternative screening strategies with different attributes, including missed diagnosis rate, overdiagnosis rate, screening result feedback efficiency, level of ophthalmologist involvement, organizational form, cost, and screening result feedback form. Conditional logit models with the stepwise selection method were used to estimate the preferences. RESULTS: Medical staff preferred high accuracy: The specificity of deep learning models should be more than 90% (odds ratio [OR]=0.61 for 10% overdiagnosis; P<.001), which was much higher than the Food and Drug Administration standards. However, accuracy was not the residents' preference. Rather, they preferred to have the doctors involved in the screening process. In addition, when compared with a fully manual diagnosis, AI technology was more favored by the medical staff (OR=2.08 for semiautomated AI model and OR=2.39 for fully automated AI model; P<.001), while the residents were in disfavor of the AI technology without doctors' supervision (OR=0.24; P<.001). CONCLUSIONS: Deep learning model under doctors' supervision is strongly recommended, and the specificity of the model should be more than 90%. In addition, digital transformation should help medical staff move away from heavy and repetitive work and spend more time on communicating with residents.

Andersen's utilization model for cataract surgical rate and empirical evidence from economically-developing areas.

BACKGROUND: Un-operated cataract is the leading cause of vision loss worldwide, responsible for 33% of visual impairment, and half of global blindness. The study aimed to build a fast evaluation method utilizing Andersen's utilization framework and identify predictors of cataract surgical rate in sub-Saharan Africa and China. METHODS: The study was a cross-over ecological epidemiology study with a total of 19 countries in sub-Saharan Africa, and 31 provinces in China. Information was extracted from public data and published studies. Linear regression and structural equation modeling with Bootstrap were used to analyze predictors of CSR and their pathways to impact in sub-Saharan Africa and China separately. RESULTS: Cataract surgical resources in sub-Saharan Africa were linearly correlated with CSR (ß = 0.74, 95% CI: 0.09, 0.91), while GDP/P didn't impact cataract surgical resources (ß = 0.29, 95% CI: - 0.12, 0.75). In China, residents' average ability to pay was confirmed as the mediator between GDP/P and CSR (p = 0.32, RMSEA = 0.07; ßCSR-paying = 0.77, 95% CI: 0.25, 0.90; ßpaying-GDP/P = 0.89, 95% CI: 0.82, 0.93). CONCLUSIONS: In sub-Saharan Africa, CSR is determined by health care provision. Local economic development may not directly influence CSR. Therefore, international assistance aimed to providing free cataract surgery directly is crucial. In China, CSR is determined principally by health care demand (ability to pay). To increase CSR in underserved areas of China, ability to pay must be enhanced through social insurance, and reduced surgical fees.

In vivo Noninvasive Imaging and Quantitative Analysis of Iris Vessels.

OBJECTIVE: To quantify the iris vessels and its circadian rhythm in normal eyes. METHODS: Fifteen healthy subjects were enrolled in this retrospective, cross-sectional study. All subjects underwent optical coherence tomography angiography (OCTA) and indocyanine green angiography (ICGA) examinations, in which 3/15 completed ICGA and OCTA at the same visit. Upon visit, consecutive OCTA scans were then obtained at the time points of the hour 3:00, 6:00, 8:00, 10:00, 12:00, 14:00, 16:00, 18:00, 20:00, 22:00, and 24:00, respectively. Vessel area density (VAD) and vessel skeleton density (VSD) were used to quantitatively describe the OCTA images of the iris vessels. RESULTS: The VAD and VSD of the iris vessels had circadian rhythm with the highest values observed at about 18:00 h and lowest at 0:00 h; the overall values were relatively stable within the 24 h. The contour analysis suggested that the iris VAD and VSD were correlated with the changes in blood pressure and inversely correlated with the changes in the intraocular pressure. CONCLUSIONS: OCTA can be used accurately for quantitative analysis of the iris vessels.

General analysis of factors influencing cataract surgery practice in Shanghai residents.

BACKGROUND: It was reported that lack of knowledge, less confidence of medical services, commute difficulties, and poor economic conditions would be the main barriers for cataract surgery practice. The influencing factors could have changed in cities with high developing speed. Shanghai is one of the biggest cities in China and the world. The purpose of the study was to explore the factors influencing cataract surgery practice in Shanghai. METHODS: This was a population-based, cross-sectional study. A total of 2342 cataract patients older than 50 years old with cataract-induced visual impairment or who had undergone cataract surgery were recruited from rural and urban areas of Shanghai. Participants accepted a face-to-face structured questionnaire. Data were collected on patient demographics, education, work, income, health insurance, awareness about cataracts disease, treatment and related medical resources and deration policy, transportation and degree of satisfaction with hospitals. RESULTS: There were 417 patients who had received cataract surgery, 404 of them supplied complete information in the questionnaire. More female subjects (64.6%) than male subjects (35.4%) accepted cataract surgery among the 404 patients. Of the patients with cataract history, 36.4% of surgery patients were equal or older than 80. More people with urban medical insurance received surgery (p = 0.036). Patients who received surgery were more satisfied with local medical service (p = 0.032). In urban area, Lower income and difficulties with commutes were related to a higher rate of surgery. CONCLUSIONS: Cataract patients with the following features were more inclined to receive surgery: female, old age, better awareness. In urban areas low income and difficult commutes did not represent barriers for cataract surgery, probably because of appropriate cataract surgery promotion policies recent years in Shanghai. In rural areas, better healthcare reimbursement policies would likely lead to a higher uptake of cataract surgery. Further cohort studies with more controls could supply stronger evidence for our viewpoint.

Different patterns of myopia prevalence and progression between internal migrant and local resident school children in Shanghai, China: a 2-year cohort study.

BACKGROUND: In 2010, there were ~ 36 million migrant children under 18 y old in China. This study compared patterns of myopia prevalence and progression between migrant and resident children. METHODS: Eight hundred forty-two migrant children from 2 migrant schools and 1081 from 2 local schools in Baoshan District, Shanghai, were randomly chosen. Baseline measurements were taken on children in grades one through four, and children in grades one and two were followed for 2 y. The children underwent comprehensive ophthalmic examinations, including cycloplegic refraction and axial length. The average time per week spent on homework and outdoor activities were investigated. RESULTS: Migrant children in grades one and two showed a lower myopia prevalence than resident children; however, from grades three to four, the prevalence accelerated and exceeded that of residents. In the follow-up, the myopia incidence did not significantly change from grades one to two in resident children but was significantly higher in grade two in migrant children. Correspondingly, for migrant children, increased progression of refraction and axial length was observed; however, it decreased in resident children. The average time spent on homework increased from grades two to three in parallel with the acceleration of myopia prevalence for migrant children; however, the time spent outdoors did not correspondingly change. CONCLUSION: The patterns of myopia prevalence and progression are different between migrant and non-migrant children. The acceleration of myopia in migrant children might be a result of a change in their environment, such as intensive education pressure.

Cohort study with 4-year follow-up of myopia and refractive parameters in primary schoolchildren in Baoshan District, Shanghai.

IMPORTANCE: Cohort studies could not only reveal associations between change of refractive components and onset/progression of myopia, but also risk factors, which is important for understanding mechanism and providing strategies. BACKGROUND: Prevalence of myopia is high in Shanghai, being reported to be 52.2% in children aged 10 years old. DESIGN: Cohort study. PARTICIPANTS: A total of 1856 students from six randomly selected primary schools in Baoshan District, Shanghai. METHODS: Children underwent comprehensive ocular measurement, including axial length (AL), corneal curvature radius and cycloplegic auto-refraction. Questionnaires about eye usage time were collected. Grade 1 students were followed for 4 years, and grade 2 and 3 students for 2 years. MAIN OUTCOME MEASURES: (i) Change of spherical equivalent (SE) and AL and (ii) risk factors for progression and incidence of myopia. RESULTS: The average 2-year progress of SE was 0.91D, 0.91D and 1.11D for grade 1, 2 and 3, respectively, and the average elongation of AL was 0.70 mm, 0.64 mm and 0.71 mm, respectively. Only parental myopia, but not near work time, near work diopter time, outdoor activity time or attending tutoring classes, was associated with myopia incidence and progression in the present population. Using baseline SE could be a simple and effective indicator for myopia prediction. CONCLUSIONS AND RELEVANCE: Incidence and progression of myopia is relatively high in schoolchildren in Shanghai compared with children of Western countries, East Asia and other parts of China. Effective strategies to control myopia prevalence are in urgent need.

Vision-related quality of life and visual outcomes from cataract surgery in patients with vision-threatening diabetic retinopathy: a prospective observational study.

BACKGROUND: To examine the benefit of cataract surgery on visual acuity and vision related quality of life in patients with stabilized vision-threatening diabetic retinopathy. METHODS: A total of 126 patients (153 eyes) who were diagnosed with cataract combined with stabilized vision-threatening diabetic retinopathy underwent phacoemulsification. Measurements included the best-corrected visual acuity (BCVA), which was converted into a weighted logarithm of the minimum angle of resolution (logMAR) and vision related quality of life (VRQoL) using the Chinese-version low vision quality of life questionnaire (CLVQOL). RESULTS: Three months after phacoemulsification, statistically significant improvements were observed in postoperative weighted logMAR BCVA (Z = -9.390 P < 0.001). In all of the participants, the CLVQOL total scores (Z = -7.995 P <0.001) and four subscale scores including general vision and lighting level (Z = -7.400 P <0.001), mobility level (Z = -6.914 P <0.001), psychological adjustment level (Z = -8.112 P <0.001) and reading, fine work and activities of daily living level (Z = -5.892 P <0.001), all improved significantly after the surgeries. Linear regression analyses indicated that the increase in CLVQOL total scores exhibited a significant correlation with the better postoperative weighted logMAR BCVA, greater gain of weighted logMAR BCVA after surgery, bilateral surgery, and longer duration of diabetic retinopathy. CONCLUSIONS: Both visual acuity and the vision related quality of life of the patients with diabetic retinopathy improved significantly after cataract surgery. Cataract surgery is an effective intervention for patients with stabilized diabetic retinopathy.

Streamlining spatial omics data analysis with Pysodb.

Advances in spatial omics technologies have improved the understanding of cellular organization in tissues, leading to the generation of complex and heterogeneous data and prompting the development of specialized tools for managing, loading and visualizing spatial omics data. The Spatial Omics Database (SODB) was established to offer a unified format for data storage and interactive visualization modules. Here we detail the use of Pysodb, a Python-based tool designed to enable the efficient exploration and loading of spatial datasets from SODB within a Python environment. We present seven case studies using Pysodb, detailing the interaction with various computational methods, ensuring reproducibility of experimental data and facilitating the integration of new data and alternative applications in SODB. The approach offers a reference for method developers by outlining label and metadata availability in representative spatial data that can be loaded by Pysodb. The tool is supplemented by a website ( https://protocols-pysodb.readthedocs.io/ ) with detailed information for benchmarking analysis, and allows method developers to focus on computational models by facilitating data processing. This protocol is designed for researchers with limited experience in computational biology. Depending on the dataset complexity, the protocol typically requires ~12 h to complete.

Study of the Properties and Modification Mechanism of SBS-Modified Asphalt by Dry Process.

SBS (styrene-butadiene-styrene block copolymer) is a thermoplastic elastomer with properties most similar to rubber. SBS asphalt modifier is mainly composed of a styrene-butadiene-styrene block copolymer with a certain amount of additives and stabilizers. SBS-modified asphalt binder has always been the most commonly used pavement material both domestically and internationally. However, conventional wet-process SBS-modified asphalt binder requires manufacturers to produce it in advance and transport it to a mixing plant for blending. This has provided an opportunity for unscrupulous businesses to reduce the amount of SBS by adding other substances, allowing inferior asphalt binder to pass inspections undetected. At the same time, conventional wet-process SBS-modified asphalt tends to undergo phase separation and experience a decline in performance as the storage time increases. However, dry-process SBS-modified asphalt can be directly added at the mixing plant, effectively addressing the issues associated with conventional wet-process SBS-modified asphalt. It also helps to reduce environmental pollution to a certain extent. This study investigates the extraction process of dry-process SBS-modified asphalt binder. It clarifies the performance and modification mechanisms of two types of dry-process SBS-modified asphalt binder at different dosages through various testing methods, including basic indicators, rheological properties, infrared spectroscopy, and fluorescence microscopy. The results indicate that due to the incorporation of oil, crosslinker, solubilizer, and other substances into dry-process SBS modifier, there is a small amount of chemical reaction with asphalt in the melting process. The high- and low-temperature properties and fatigue properties of the two dry-process SBS-modified asphalt binders at a 7% dosage are close to wet SBS-modified asphalt binder at a 5% dosage.

RETFound-enhanced community-based fundus disease screening: real-world evidence and decision curve analysis.

Visual impairments and blindness are major public health concerns globally. Effective eye disease screening aided by artificial intelligence (AI) is a promising countermeasure, although it is challenged by practical constraints such as poor image quality in community screening. The recently developed ophthalmic foundation model RETFound has shown higher accuracy in retinal image recognition tasks. This study developed an RETFound-enhanced deep learning (DL) model for multiple-eye disease screening using real-world images from community screenings. Our results revealed that our DL model improved the sensitivity and specificity by over 15% compared with commercial models. Our model also shows better generalisation ability than AI models developed using traditional processes. Additionally, decision curve analysis underscores the higher net benefit of employing our model in both urban and rural settings in China. These findings indicate that the RETFound-enhanced DL model can achieve a higher net benefit in community-based screening, advocating its adoption in low- and middle-income countries to address global eye health challenges.

Corrigendum: Establishment of noncycloplegic methods for screening myopia and pre-myopia in preschool children.

[This corrects the article DOI: 10.3389/fmed.2023.1291387.].

Complete spatially resolved gene expression is not necessary for identifying spatial domains.

Spatially resolved transcriptomics (SRT) technologies have revolutionized the study of tissue organization. We introduce a graph convolutional network with an attention and positive emphasis mechanism, termed BINARY, relying exclusively on binarized SRT data to accurately delineate spatial domains. BINARY outperforms existing methods across various SRT data types while using significantly less input information. Our study suggests that precise gene expression quantification may not always be essential, inspiring further exploration of the broader applications of spatially resolved binarized gene expression data.

Sympathetic nervous system activity is associated with choroidal thickness and axial length in school-aged children.

BACKGROUND/AIMS: We aim to explore the effect of sympathetic nervous system (SNS) on choroid thickness (ChT) and axial length (AL). METHODS: Students of grade 2 and 3 from a primary school were included and followed for 1 year. Visual acuity, refraction, AL and ChT were measured. Morning urine samples were collected for determining SNS activity by analysing concentrations of epinephrine, norepinephrine and dopamine using the liquid chromatography-tandem mass spectrometry. The most important factor (factor 1) was calculated using factor analysis to comprehensively indicate the SNS activity. RESULTS: A total of 273 students were included, with an average age of 7.77±0.69 years, and 150 (54.95%) were boys. Every 1 µg/L increase in epinephrine is associated with 1.60 µm (95% CI 0.30 to 2.90, p=0.02) decrease in average ChT. Every 1 µg/L increase in norepinephrine is associated with 0.53 µm (95% CI 0.08 to 0.98, p=0.02) decrease in the ChT in inner-superior region. The factor 1 was negatively correlated with the ChT in the superior regions. Every 1 µg/L increase in norepinephrine was associated with 0.002 mm (95% CI 0.0004 to 0.004, p=0.016) quicker AL elongation. The factor 1 was positively correlated with AL elongation (coefficient=0.037, 95% CI 0.005 to 0.070, p=0.023). CONCLUSIONS: We hypothesised that chronic stress characterised by elevated level of the SNS, was associated with significant increase in AL elongation, probably through thinning of the choroid.

Sparse and Hierarchical Transformer for Survival Analysis on Whole Slide Images.

The Transformer-based methods provide a good opportunity for modeling the global context of gigapixel whole slide image (WSI), however, there are still two main problems in applying Transformer to WSI-based survival analysis task. First, the training data for survival analysis is limited, which makes the model prone to overfitting. This problem is even worse for Transformer-based models which require large-scale data to train. Second, WSI is of extremely high resolution (up to 150,000 x 150,000 pixels) and is typically organized as a multi-resolution pyramid. Vanilla Transformer cannot model the hierarchical structure of WSI (such as patch cluster-level relationships), which makes it incapable of learning hierarchical WSI representation. To address these problems, in this paper, we propose a novel Sparse and Hierarchical Transformer (SH-Transformer) for survival analysis. Specifically, we introduce sparse self-attention to alleviate the overfitting problem, and propose a hierarchical Transformer structure to learn the hierarchical WSI representation. Experimental results based on three WSI datasets show that the proposed framework outperforms the state-of-the-art methods.

Artificial Intelligence in Community-Based Diabetic Retinopathy Telemedicine Screening in Urban China: Cost-effectiveness and Cost-Utility Analyses With Real-world Data.

BACKGROUND: Community-based telemedicine screening for diabetic retinopathy (DR) has been highly recommended worldwide. However, evidence from low- and middle-income countries (LMICs) on the choice between artificial intelligence (AI)-based and manual grading-based telemedicine screening is inadequate for policy making. OBJECTIVE: The aim of this study was to test whether the AI model is more worthwhile than manual grading in community-based telemedicine screening for DR in the context of labor costs in urban China. METHODS: We conducted cost-effectiveness and cost-utility analyses by using decision-analytic Markov models with 30 one-year cycles from a societal perspective to compare the cost, effectiveness, and utility of 2 scenarios in telemedicine screening for DR: manual grading and an AI model. Sensitivity analyses were performed. Real-world data were obtained mainly from the Shanghai Digital Eye Disease Screening Program. The main outcomes were the incremental cost-effectiveness ratio (ICER) and the incremental cost-utility ratio (ICUR). The ICUR thresholds were set as 1 and 3 times the local gross domestic product per capita. RESULTS: The total expected costs for a 65-year-old resident were US $3182.50 and US $3265.40, while the total expected years without blindness were 9.80 years and 9.83 years, and the utilities were 6.748 quality-adjusted life years (QALYs) and 6.753 QALYs in the AI model and manual grading, respectively. The ICER for the AI-assisted model was US $2553.39 per year without blindness, and the ICUR was US $15,216.96 per QALY, which indicated that AI-assisted model was not cost-effective. The sensitivity analysis suggested that if there is an increase in compliance with referrals after the adoption of AI by 7.5%, an increase in on-site screening costs in manual grading by 50%, or a decrease in on-site screening costs in the AI model by 50%, then the AI model could be the dominant strategy. CONCLUSIONS: Our study may provide a reference for policy making in planning community-based telemedicine screening for DR in LMICs. Our findings indicate that unless the referral compliance of patients with suspected DR increases, the adoption of the AI model may not improve the value of telemedicine screening compared to that of manual grading in LMICs. The main reason is that in the context of the low labor costs in LMICs, the direct health care costs saved by replacing manual grading with AI are less, and the screening effectiveness (QALYs and years without blindness) decreases. Our study suggests that the magnitude of the value generated by this technology replacement depends primarily on 2 aspects. The first is the extent of direct health care costs reduced by AI, and the second is the change in health care service utilization caused by AI. Therefore, our research can also provide analytical ideas for other health care sectors in their decision to use AI.