The Ocular Epidemiology Research Group, led by Prof Cheng Ching-Yu, aims to develop and facilitate collaborative and translational research by integrating big data, omics, and machine learning analytics into large scale population-based studies. The strategic goal of our group is to improve the eye health of populations in Singapore and globally, and to become one of the leading international centres for population science and digital health of eye diseases by:
using state-of-art data analytics and artificial intelligence techniques,
establishing collaborative research platform, and
nurturing the next generation of researchers and scientists.
Our Vision
Build a world-leading research programme focusing specifically on the epidemiology and population health of major eye diseases in Asia.
Provide a one-stop “data portal” and information source on the population health of Asian eye diseases.
Foster international collaborations with other population health and ophthalmic institutes in Asia and worldwide.
Our Aims
To document the prevalence, incidence, risk factors and public health significance of blinding eye diseases in Singapore and Asia by conducting large scale epidemiological studies under the umbrella of ‘Singapore Epidemiology of Eye Diseases (SEED)’ Programme.
To support research initiatives by developing and maintaining population data, biospecimen, genomic resources, images, and analytics methodologies.
To bridge the gap between population health and clinical applications by leveraging existing biospecimen resources in discovery or validation of biomarkers for risk stratification, and developing novel screening modalities based on deep learning technologies.
To provide research expertise, training and consultation to other researchers and ophthalmic institutions in Singapore, the Asia-Pacific region and globally.
The Ocular Epidemiology Research Group brings together innovative population health research and cutting-edge technology, such as a next-generation sequencing and artificial intelligence, with a focus on our theme-oriented strategy. Our group has multi-disciplinary expertise in all aspects of clinical and epidemiological research. We mainly research and develop the following areas:
Translational Population Health
Big Data Analytics and Artificial Intelligence
Omics and Genetic Epidemiology
Infrastructure of Population Science
Cohort
Age range
Ethnicity
Number
Data Collected
Singapore Malay Eye Study (SiMES)
40-80 years
Malays
3280
Prevalence, risk factors, and impact of visual impairment and major eye diseases in Singaporean Malays
Singapore Indian Eye Study (SINDI)
Indians
3400
Prevalence, risk factors, and impact of visual impairment and major eye diseases in Singaporean Indians
Singapore Chinese Eye Study (SCES)
Chinese
3353
Prevalence, risk factors, and impact of visual impairment and major eye diseases in Singaporean Chinese
The Singapore Prospective Study Programme (SP2) Ancillary study
24-95 years
Chinese, Malays and Indians
5000
Prevalence, environmental and genetic risk factors, and impacts of cardiovascular and metabolic diseases (e.g., hypertension, dyslipidemia, obesity, and diabetes mellitus)
Polyclinic Study
50+ years
2000
Ocular and image biomarkers for glaucoma
Cohort Highlights
Singapore Singapore Epidemiology of Eye Diseases (SEED) Study
The SEED study is a multi-ethnic longitudinal population-based study, comprised of over 10,000 adult participants aged 40 years or older in Singapore. It included three large population-based studies: the Singapore Malay Eye Study (SiMES), the Singapore Indian Eye Study (SINDI), and the Singapore Chinese Eye Study (SCES), with a focus on studying major eye diseases, including diabetic retinopathy, age-related macular degeneration, glaucoma, refractive errors and cataract. As one of the largest epidemiological databases and biobanks for eye diseases globally, SEED data has been widely used by national and international agencies to help guide public policy decisions on screening and early detection of age-related eye diseases. Furthermore, SEED findings have also contributed to other epidemiological areas such as machine/deep learning and genetic discoveries. More detailed information regarding SEED study can be found here.
Asian Epidemiology of Eye Diseases (AEEC) Consortium
The Asian Eye Epidemiology Consortium (AEEC) is a collaborative network of population-based studies performed across Asia with the overall aim of developing large datasets to provide deeper insights on the trends and associated risk factors of major age-related eye diseases among Asians. The AEEC network consist of over 40 population-based studies originating from at least 10 different Asian countries. Currently, data from the AEEC network has provided key insights into major age-related eye diseases in Asians, such as geographic atrophy, primary open-angle glaucoma, myopia, and diabetic retinopathy. We welcome new population studies to join the consortium.
Project
Principal Investigator
Period
Artificial Intelligence for Functional VIsion Screening Using Retinal Imaging (AVIRI)
Dr Tham Yih Chung
2019 - 2021
Community-based Screening for Pathological Visual Impairment among Elderly Residents using Artificial-Intelligence Integrated Retinal Imaging
2019 - 2022
From Machine to Machine-developing a Deep Learning Algorithm for Quantification of Ocular Traits based on Retinal Photographs
Dr Tyler Hyungtaek Rim
2020 - 2021
Implementation of Community-based Elderly Health Care for Eye and Systemic Diseases Using Automated Screening
2020 - 2022
SERI-based Machine Learning and AI Talent (SMAT) Programme
The Retina as a Window to Vascular and Neurological Disorders
Prof Wong Tien Yin
2021 - 2022
Digital and Precision Community Screening Platform for Ageing Diseases: Vision, Metabolism and Heart
Prof Cheng Ching-Yu
2021 - 2025
Project Highlights
New AI-assisted Vision Screening Model for Community
Visual impairment is a major public health problem, associated with reduced quality of life and increased risk of frailty and mortality. Globally in 2020, an estimated 553 million people had visual impairment and 43 million were blind. 40% of visual impairment is related to refractive error that requires the provision of spectacles in community settings; however, the remaining 60% of cases cannot be corrected with spectacles and require assessment, diagnosis, treatment, and possibly surgery in eye-care settings led by ophthalmologists. These 60% of people with visual impairment can be referred to as having disease-related visual impairment (i.e., substantial loss of vision caused by an eye disease, and unrelated to refractive error). If detected early, these conditions can be treated, thus preventing or slowing development of vision loss. We developed a single-modality, retinal photograph-based deep learning algorithm (termed AVIRI) to detect disease-related visual impairment, using a total of 15,175 eyes from a multi-ethnic Asian population-based eye study. We also did independent validation of the algorithm using datasets of eyes from three other population-based studies and two clinic-based studies (total of 16,963 eyes), which generally showed that the algorithm had optimal performance. This is the first study to show the use of a single-modality deep learning algorithm, using only a single macular-centred retinal photograph, for identification. The unique design of this algorithm enables it to potentially be used as an efficient automated referral tool in community screening. Moving forward, there are plans to perform real-world validation to enable translation of this innovation for community screening use. Our work was published in Lancet Digital Health and was featured as Editor’s pick. More details can be found here.
Deep-learning-based Cardiovascular Risk Stratification using Coronary Artery Calcium Scores Predicted from Retinal Photographs
Cardiovascular disease is the leading cause of death worldwide. The retina is the only organ that allows direct, non-invasive, in-vivo visualisation of the microvasculature and neural tissues. In recent decades, our understanding of retina-systemic relationships has relied on classic epidemiological studies based on observable, human-defined retinal features (e.g., retinopathy or retinal vascular calibre). The potential discovery of unobservable retinal features associated with systemic diseases has been enhanced by advances in artificial intelligence technology, specifically deep learning. Coronary artery calcium (CAC) is a preclinical marker of atherosclerosis and is strongly associated with risk of clinical cardiovascular disease. We have extended this concept of retina-systemic relationships and hypothesise that retinal photograph-based deep learning can also predict CAC score, and this retinal-predicted score (termed “RetiCAC) can also be used as a risk stratification tool for cardiovascular events. In our study, similar to the current CT-measured CAC stratification system, the relative risk of cardiovascular disease events showed a dose-response association across the three risk strata. Overall, the proposed new stratification system based on RetiCAC score showed comparable performance in predicting cardiovascular disease events compared with conventional CT-measured CAC score. Thus, retinal photography could potentially be adopted as a relatively simple and non-radiation imaging modality for cardiovascular disease risk classification. Our work was published in Lancet Digital Health.
A Deep Learning Algorithm to Detect Chronic Kidney Disease from Retinal Photographs in Community-based Populations
Chronic kidney disease (CKD) is a major health condition associated with significant morbidity, cardiovascular disease and mortality. Screening for CKD is challenging in community and primary care settings, even in high-income countries because of the need to obtain serum levels of creatinine, or testing urine for protein. The retina being accessible to non-invasive imaging and retinal changes have been shown to provide information on systemic vascular and metabolic diseases, we developed an artificial intelligence deep learning algorithm (DLA) to detect CKD from retinal images, which may add to existing CKD screening strategies. We developed and validated the DLA utilising retinal images and data from the Singapore Epidemiology of Eye Diseases (SEED) and externally validated the DLA in two independent datasets in Singapore (SP2) and China (BES). We developed an ‘Image-only’ model based on macula-centred images from both eyes. For comparison, we also developed a ‘Risk-factor’ model based on key risk factors including age, sex, ethnicity, diabetes and hypertension. The image-only DLA showed an AUC of 0.911 in internal validation and 0.733 and 0.835 in external tests sets. Corresponding estimates for the risk-factor model were 0.916, 0.829 and 0.887. The image-only DLA and risk factor DLAs achieve high AUCs in SEED internal validation and modest to good AUCs in external test sets. Our findings show that for CKD detection, a retinal-image only DLA is similar to information from a classic risk-factor model and support the potential of a retinal-imaged based DLA to be adopted for first stage CKD screening before confirmation by serum creatinine. Our work was published in Lancet Digital Health. More details can be found here.
We have established Singapore as a leading hub of ophthalmic epidemiology, clinical and genetic research in Asia with a particular focus on diseases that are prevalent in this region. Data collected from our group has been used widely by national and international agencies (e.g. the Ministry of Health (MOH) Singapore, the World Health Organization, the Global Burden of Disease programme, etc.) and clinical guidelines (e.g. 2014 MOH Diabetes Guidelines, 2016 Asia Pacific Glaucoma Guidelines, 2016 American Diabetes Association Guidelines, 2017 International Council of Ophthalmology Diabetic Eye Care Guidelines). The data has been used to provide estimates of eye disease burden to set up the national DR screening. It also assists both the MOH’s planning for healthcare manpower (future ophthalmology and optometry manpower) and the College of Ophthalmology’s planning for the National Ophthalmology Road Map 2030 plan in Singapore.
In addition, we also provide research expertise, resource, and consultation for other research centres, hospitals and ophthalmic institutions in Singapore and Asia. We conduct training programmes for clinicians, research fellows and graduate students interested in big data analytics, digital health, and ophthalmic epidemiology.
The list below is selected from more than 600 publications.
Head
Deputy Head
Assoc Prof Charumathi Sabanayagam
Investigators & Fellows
Dr Simon Nusinovici
Dr Sahil Thakur
Research Associate / Officer
Dr Chen Yanyan
Dr Shivani Majithia
Soh Zhi Da
Peng Qingsheng
Quek Ten Cheer
Jocelyn Goh Hui Lin
He Feng
Zann Lee Yan Shin
Data Science Team
Dr Marco Yu
Chee Miao Li
Crystal Chong
Mihir Deshmukh
Dr Fan Qiao
Research Clinic Team
Teo Cong Ling
Binu Thapa
Kaeley Koh Kai Hui
Sarah Tan Shwu Huey
Rachel Marjorie Tseng Wei Wen
Cheok Kaa Ming
Gao Fei
Rosesita Binte Shaikh
Manivannan Udayaraj
Shernisee Chia
Biobank Laboratory
Chee Miao Ling
Chia Boon Jun
Research Administration
Ho Kee Ka
Riswana Banu
SNEC Clinical Team
Prof Gemmy Cheung Chui Ming
Assoc Prof Gavin Tan Siew Wei
Dr Kelvin Teo Yi Chong
Assoc Prof Danny Cheung Ning
Clin Assoc Prof Anna Tan Cheng Sim
Dr Beau James Fenner
Dr Nicholas Tan Yi Qiang
Dr Lim Sing Hui
Dr Stanley Poh Shuoh Jieh
Dr Teo Zhen Ling
Dr Debra Quek Qiao Yun
Dr Ryan Lee
SERI Research Team
Prof Ecosse Lamoureux
Prof Saw Seang Mei
Prof Leopold Schmetterer
Prof Louis Tong
Assoc Prof Zhou Lei
Dr Jacqueline Chua Yu Min
Dr Ryan Mann
Dr Preeti Gupta
Subscribe to our mailing list to get the updates to your email inbox...
