AI IN Healthcare

Working closely with the clinical team from the Singapore Eye Research Institute (SERI) under SingHealth, the Institute of High Performance Computing (IHPC) has developed multiple AI models for eye diseases detection from fundus photos. These include papilledema, visually significant cataract, disease-related visual impairment, myopic macular degeneration and high myopia. The research works have been published in New England Journal of Medicine, Nature Ageing, The Lancet Digital Health and Nature Medicine.

A*STAR’s Institute for Infocomm Research (I2R) and IHPC also worked with Tan Tock Seng Hospital (TTSH) to develop an AI solution to detect Covid-19 using chest X-ray images. This system was trialled at NCID and awarded the National Health Innovation and Productivity Medal 2021 as well as Singapore Computer Society IT Leader Award 2021.

Fig 2: Dr Ting Yong Han, consultant in diagnostic radiology (clinical) at TTSH, looking at a chest X-ray of a Covid-19 patient that was flagged by the AI tool, RadiLogic, which can analyse an image within three seconds. Photo credit: The Straits Times.

In addition, together with Diagnostic Development Hub (DxD), I2R developed an AI-based software to reduce human error and augment productivity in the use of the Resolute kit - a PCR-based diagnostic kit for the detection of Covid-19. This technology was eventually licensed and used by Advanced Medtech Solutions as part of their optimised workflow.

The Bioinformatics Institute (BII) also performed a series of digital pathology projects with National Cancer Centre Singapore (NCCS), Singapore General Hospital (SGH) and TTSH, developing AI tools that can read and diagnose high resolution digital pathology slides for prostate, lung, and other types of cancers.

Doctor Covid
In support of Singapore’s COVID-19 management, SingHealth and IHPC, with support from the Integrated Health Information Systems (IHiS), have co-developed a cloud-based telemedicine platform known as “Doctor Covid” to improve care for COVID-19 patients at community care facilities. Hosted on the Telegram mobile application as a communication channel, Doctor Covid boasts a variety of features in the patients’ native languages to facilitate effective communication, active engagement and interaction, and to support the medical and operations teams in close monitoring of their health and well-being.

A*STAR’s clinical partners deployed the first system in 3 Community Care Facilities, assisting the management of more than 4,000 COVID patients with almost 1 million messages sent.  The system was integrated into the SGH covid virtual ward project to serve COVID patients recovering at home. Recent estimates (as of March 2023) from SGH have indicated the saving of more than 7,500 bed days, translating into more than SGD $7.75 million cost savings, in addition to a reduction of 194-man days of clinical staff. IHPC continues to develop this platform, with the intention to incorporate more advanced solutions for telehealth such as vital sign monitoring and symptoms identification to assist doctors to triage and diagnose.

Diabetes Clinic of the Future
SingHealth and A*STAR established a “Diabetes Clinic of the Future” by combining patient data and clinical expertise from SingHealth, with medical technology, data analytics, artificial intelligence and advanced imaging capabilities from A*STAR.

The Clinic will utilise smart technologies to improve diabetes care across the disease continuum- from risk prediction to the monitoring- treatment and management of diabetes. Technologies introduced could include mobile applications or remote monitoring technologies which can help to track the nutritional intake- glucose levels- and fitness levels of at-risk (pre-diabetic and diabetic) populations- to help them better manage and take charge of their health. The Clinic is also looking at nurse-patient conversations, and is engaging with Changi General Hospital on developing a conversation agent as part of their home-care programme.

Drug discovery & development is a complex and lengthy endeavour with high levels of uncertainty; a multitude of approaches is necessary to rigorously investigate and uncover new targets and drugs. Given the breadth and scope involved as well as the need to integrate interdisciplinary approaches, A*STAR is adopting artificial intelligence and advancing its computational capabilities in diverse ways to accelerate drug discovery efforts across the entire value chain. 

AI tools can also be used to analyse large volumes of biological and chemical data to identify new drug candidates. IHPC has designed advanced deep learning-based algorithms that can accurately predict the binding affinity of small molecules to target proteins. 

By working together, researchers from different institutes can pool domain expertise and resources to tackle drug discovery and development from a range of perspectives. For instance, Experimental Drug Development Centre (EDDC) is working with GIS to accelerate the development of new RNA-targeting therapeutics. The platform is also supporting BII and IHPC in designing drug candidates to target newly discovered allosteric or cryptic binding sites on proteins, which could lead to better treatments with minimal side effects.

Coronary artery disease, a blockage of the blood vessels, is a leading cause of cardiac mortality in Singapore and worldwide, contributing to 19% of deaths in Singapore. An AI platform called APOLLO (AI-driven national Platform for CT cOronary angiography for clinical and industrial applicatiOns) was developed and prototyped by A*STAR, National Heart Centre Singapore (NHCS), National University Health System (NUHS) and Tan Tock Seng Hospital (TTSH) to help doctors to provide a rapid and comprehensive diagnosis of coronary artery diseases. 

The APOLLO platform integrates several AI toolkits that analyse Computed Tomography Coronary Angiography (CTCA) scans into clinical workflow. By integrating AI toolkits with diagnostic and prognostic capabilities, the “one-stop” platform enables them to work functionally together to provide automated anonymization, reporting, calcium scoring, epicardial fat, coronary stenosis and plaque quantification. The AI tools and systems that are integrated on the APOLLO platform were developed by researchers from BII and I2R.

Precision medicine aims to tailor healthcare interventions to individual patients based on their unique characteristics, including genetics, lifestyle, and environmental factors. Genome Institute of Singapore (GIS) is revolutionising healthcare with AI-driven precision medicine by combining advanced AI algorithms with comprehensive patient data, including genomics, clinical records and lifestyle factors. By uncovering hidden patterns and correlations, GIS empowers healthcare professionals to deliver personalised treatments tailored to individual needs, ensuring optimal outcomes and minimising the burden of disease. 

National Precision Medicine (NPM) Programme

The SG10K_Health project is a multi-institutional initiative developing an ‘all-of-Singapore' coordinated approach to precision medicine. At present, the Asian population is severely under-represented in the public genotypic databases. The current lack of large-scale control databases of Asian-specific genetic variation linked to clinical phenotypes is a significant barrier to the conduct of precision medicine in Asia, to avoid mis-diagnosis and overtreatment due to the mistaken identification of pathogenic variants.

The SG10K_Health project aims to empower biomedical and genetic studies of Singapore and Asian-centric diseases by: 1) building local infrastructure and deep capabilities to generate, analyse and store human genetic data at population scale, in a safe, secure and rapid manner, 2) generating a diverse and rich control dataset of Asian populations for genetic association study of diseases, and 3) developing advanced analytical tools for genetic variants interpretation to derive disease risk predictions and identify clinically actionable variants. Currently, the SG10K_Health data is linked to research traits (e.g., height, weight, blood pressure) and in the future will be linked to clinical records, subject to participant consent.