Background. Deep learning has been widely used for glaucoma diagnosis. However, there is no clinically validated algorithm for glaucoma incidence and progression prediction. This study aims to develop a clinically feasible deep-learning system for predicting and stratifying the risk of glaucoma onset and progression based on color fundus photographs (CFPs), with clinical validation of performance in external population cohorts. Methods. We established data sets of CFPs and visual fields collected from longitudinal cohorts. The mean follow-up duration was 3 to 5 years across the data sets. Artificial intelligence (AI) models were developed to predict future glaucoma incidence and progression based on the CFPs of 17,497 eyes in 9346 patients. The area under the receiver operating characteristic (AUROC) curve, sensitivity, and specificity of the AI models were calculated with reference to the labels provided by experienced ophthalmologists. Incidence and progression of glaucoma were determined based on longitudinal CFP images or visual fields, respectively. Results. The AI model to predict glaucoma incidence achieved an AUROC of 0.90 (0.81–0.99) in the validation set and demonstrated good generalizability, with AUROCs of 0.89 (0.83–0.95) and 0.88 (0.79–0.97) in external test sets 1 and 2, respectively. The AI model to predict glaucoma progression achieved an AUROC of 0.91 (0.88–0.94) in the validation set, and also demonstrated outstanding predictive performance with AUROCs of 0.87 (0.81–0.92) and 0.88 (0.83–0.94) in external test sets 1 and 2, respectively. Conclusion. Our study demonstrates the feasibility of deep-learning algorithms in the early detection and prediction of glaucoma progression. FUNDING. National Natural Science Foundation of China (NSFC); the High-level Hospital Construction Project, Zhongshan Ophthalmic Center, Sun Yat-sen University; the Science and Technology Program of Guangzhou, China (2021), the Science and Technology Development Fund (FDCT) of Macau, and FDCT-NSFC.

Sex differences are widely observed under various circumstances ranging from physiological processes to therapeutic responses, and a myriad of sex-biased genes have been identified. In recent years, transcriptomic datasets of microRNAs (miRNAs), an important class of non-coding RNAs, become increasingly accessible. However, comprehensive analysis of sex difference in miRNA expression has not been performed. Here, we identified the differentially-expressed miRNAs between males and females by examining the transcriptomic datasets available in public databases and conducted a systemic analysis of their biological characteristics. Consequently, we identified 73 female-biased miRNAs (FmiRs) and 163 male-biased miRNAs (MmiRs) across four tissues including brain, colorectal mucosa, peripheral blood, and cord blood. Our results suggest that compared to FmiRs, MmiRs tend to be clustered in the human genome and exhibit higher evolutionary rate, higher expression tissue specificity, and lower disease spectrum width. In addition, functional enrichment analysis of miRNAs show that FmiR genes are significantly associated with metabolism process and cell cycle process, whereas MmiR genes tend to be enriched for functions like histone modification and circadian rhythm. In all, the identification and analysis of sex-biased miRNAs together could provide new insights into the biological differences between females and males and facilitate the exploration of sex-biased disease susceptibility and therapy.

Suboptimal health status (SHS) is a physical state between health and disease, characterized by the perception of health complaints, general weakness, chronic fatigue and low energy levels. SHS is proposed by the ancient concept of traditional Chinese medicine (TCM) from the perspective of preservative, predictive and personalized (precision) medicine. We previously created the suboptimal health status questionnaire 25 (SHSQ-25), a novel instrument to measure SHS, validated in various populations. SHSQ-25 thus affords a window of opportunity for early detection and intervention, contributing to the reduction of chronic disease burdens. To investigate the causative effect of SHS in non-communicable chronic diseases (NCD), we initiated the China suboptimal health cohort study (COACS), a longitudinal study starting from 2013. Phase I of the study involved a cross-sectional survey aimed at identifying the risk/protective factors associated with SHS; and Phase II: a longitudinal yearly follow-up study investigating how SHS contributes to the incidence and pattern of NCD. (1) Cross-sectional survey: in total, 4313 participants (53.8 % women) aged from 18 to 65 years were included in the cohort. The prevalence of SHS was 9.0 % using SHS score of 35 as threshold. Women showed a significantly higher prevalence of SHS (10.6 % in the female vs. 7.2 % in the male, P < 0.001). Risk factors for chronic diseases such as socioeconomic status, marital status, highest education completed, physical activity, salt intake, blood pressure and triglycerides differed significantly between subjects of SHS (SHS score ≥35) and those of ideal health (SHS score <35). (2) Follow up: the primary and secondary outcomes will be monitored from 2015 to 2024. The sex-specific difference in prevalence of SHS might partly explain the gender difference of incidence of certain chronic diseases. The COACS will enable a thorough characterization of SHS and establish a cohort that will be used for longitudinal analyses of the interaction between the genetic, lifestyle and environmental factors that contribute to the onset and etiology of targeted chronic diseases. The study together with the designed prospective cohort provides a chance to characterize and evaluate the effect of SHS systemically, and it thus generates an unprecedented opportunity for the early detection and prevention of chronic disease.

Glycosylation significantly affects protein structure and function and thus participates in multiple physiologic and pathologic processes. Studies demonstrated that immunoglobulin G (IgG) N-glycosylation associates with the risk factors of ischemic stroke (IS), such as aging, obesity, dyslipidemia, type 2 diabetes, and hypertension. The study aimed to investigate the association between IgG N-glycosylation and IS in a Chinese population. IgG glycome composition in patients with IS (n = 78) and cerebral arterial stenosis (CAS) (n = 75) and controls (n = 77) were analyzed by ultra-performance liquid chromatography. Eleven initial glycans and 10 derived glycans in IgG glycome representing galactosylation, sialylation, and bisecting GlcNAc significantly differed between IS patients and CAS and healthy controls after controlling for gender, age, obesity, diabetes, hypertension, and dyslipidemia. Logistic regression models incorporating IgG glycan traits were able to distinguish IS from CAS (area under receiver–operator characteristic curves (AUC), 0.802; 95% confidence interval (CI), 0.732–0.872) and controls (AUC, 0.740; 95% CI, 0.661–0.819). The canonical correlation analysis indicated that initial N-glycan structures are significantly correlated with inflammation markers (r = 0.566, p < 0.001). Our findings indicated that loss of galactose and sialic acid, as well as addition of bisecting GlcNAc, might involve in pro- or anti-inflammatory IgG functionality and further contribute to the pathogenesis of IS. IgG glycan profiles may be developed as clinical useful biomarkers for chronic disease in the future.