Building an artificial intelligence and digital ecosystem: a smart hospital’s data-driven path to healthcare excellence

Abstract

Hospitals worldwide recognise the importance of data and digital transformation in healthcare. We traced a smart hospital’s data-driven journey to build an artificial intelligence and digital ecosystem (AIDE) to achieve healthcare excellence. We measured the impact of data and digital transformation on patient care and hospital operations, identifying key success factors, challenges, and opportunities. The use of data analytics and data science, robotic process automation, AI, cloud computing, Medical Internet of Things and robotics were stand-out areas for a hospital’s data-driven journey. In the future, the adoption of a robust AI governance framework, enterprise risk management system, AI assurance and AI literacy are critical for success. Hospitals must adopt a digital-ready, digital-first strategy to build a thriving healthcare system and innovate care for tomorrow.

INTRODUCTION

Healthcare systems worldwide are facing increasingly complex challenges in the areas of lack of manpower and rising healthcare costs. These challenges are even more apparent in Singapore, given the ageing population and significant increase in healthcare spending. Singapore will reach super-aged status by 2026. This is when the proportion of the population aged 65 years and above reaches the 21% mark. By 2030, around one in four Singaporeans will be aged 65 years and above, up from one in ten in 2010. Expenditure by Singapore’s Ministry of Health (MOH) is expected to surge by SGD 2.9 billion or 16.3% year on year.[1,2,3,4]

Yet at the same time, there are opportunities in the use of data science and artificial intelligence (AI) to improve productivity and optimise healthcare processes and systems. The shift towards Healthier SG, population health and capitation model aims to ‘bend the curve’ by reducing healthcare cost expenditures and improving the health outcomes of the population. Artificial intelligence and data science have been reported to achieve a significant reduction in manpower and improve hospital efficiency. Singapore’s Minister of Health, Mr Ong Ye Kung shared, “The advent of artificial intelligence further promises major breakthroughs in healthcare. This has presented us with an unprecedented opportunity to transform healthcare for the better”. The MOH will make a centralised push to scale AI use cases into national projects. This will include leveraging Generative AI for routine documentation of medical records, which will save considerable time for healthcare professionals. Additionally, AI imaging technology will be deployed across public hospitals for more accurate, error-free diagnoses. Health Empowerment through Advanced Learning and Intelligent eXchange (HEALIX), a cloud-based data infrastructure, will also be used to train AI and machine learning (ML) tools.[5,6]

Several publications have highlighted the use of AI and data science to improve the productivity and efficiency of hospital operations in both inpatient and outpatient settings.[7,8,9,10,11,12,13] To ensure successful adoption, the hospital’s digital transformation and infrastructure must be well established. This paper reviews the data-driven journey of a 1,000-bed smart hospital and public healthcare institution (PHI), highlighting the key strategies that enabled the successful establishment of an AI and digital ecosystem (AIDE) to achieve healthcare excellence.

BUILDING STRONG FOUNDATIONS IN DATA SCIENCE

As Changi General Hospital (CGH) evolved into a data-centric organisation, data governance, conformance to regulatory requirements on data protection and analytics were recognised as strategically important drivers for organisational growth.[14,15] A specialised department, Data Management and Informatics (DMI), was established in 2014 to manage data extraction, analysis and data sharing through a data clearing house and governance framework. It provided statistical reporting services to fulfil MOH, healthcare cluster and regulatory requirements, and supported senior management and key stakeholders in making informed decisions using business intelligence and analytical tools.

The coronavirus disease 2019 (COVID-19) pandemic presented both challenges and opportunities across clinical, operational and administrative domains to ensure clinical quality and patient safety. Lessons from the pandemic included the need to improve healthcare data science and automation capabilities, such as predictive and prescriptive analytics, AI, ML and robotic process automation (RPA). These capabilities have the potential to improve patient safety and achieve productivity gains, including full-time equivalent (FTE) savings and greater hospital efficiency.

Recognising the importance of AI and data science, we pivoted DMI in 2024 to form the new Data Science and Intelligence (DSI) team, including the establishment of an AI office. This reflects their expanded roles in strategic data and AI governance, decision support analytics and advanced data science. The DSI team is responsible for these functions across domains such as healthcare intelligence, AI, ML, deep learning (DL), data and AI literacy, RPA, data and AI governance, trusted third-party-based anonymisation and data management [Figure 1].

Figure 1.

Role of the Data Science and Intelligence team at Changi General Hospital (CGH). CDP: career development programme, DL: deep learning, IPMOF: Institute Program Management Oversight Office, ML: machine learning, MOHH: Ministry of Health Holdings, NUS: National University of Singapore, PDPA: Personal Data Protection Act, PQ: parliamentary query, innovation and enterprise, RPA: robotic process automation, SHS: SingHealth Services, SMU: Singapore Management University, SUTD MM: Singapore University of Technology and Design ModularMasters, TP: Temasek Polytechnic, UAM: user acess matrix, WAU: weighted activity unit.

IMPLEMENTATION OF ROBOTIC PROCESS AUTOMATION

Robotic process automation in healthcare can significantly enhance productivity, reduce errors and care gaps and improve process efficiency. It automates repetitive, rule-based tasks, freeing up staff to focus on high-value tasks. Our RPA journey began in 2017 through the development of automated scripts for data mining and reporting using open-source tools. Over the years, we have successfully implemented RPA to improve operational efficiency, including automating the tracking of medication orders, improving dispensing times and reducing medication errors, processing discharge summaries and optimising the e-financial counselling process. During the COVID-19 pandemic, we developed a novel automated contact tracing algorithm[16] enabled by RPA scripts that streamlined outpatient registrations, optimised appointment scheduling for safe-distancing compliance, improved laboratory processes and automated statutory case reporting. Recently, we successfully automated lab processes at the Shimadzu–CGH Clinomics Centre (SC3), which translated to significant productivity gains (approximately 0.4 FTE annualised with the opportunity to scale up the automation for other laboratory diagnostic tests). These success stories demonstrate the potential of RPA to transform healthcare, augment safety and improve enterprise productivity.

BUILDING STRONG DATA GOVERNANCE SYSTEMS

As AI continues to transform healthcare, ensuring its responsible development and deployment is crucial. The CGH team developed an AI Governance Framework that integrates regulatory requirements, ethical considerations and technical standards. The framework is aligned with MOH’s Artificial Intelligence in Healthcare Guidelines,[17] the European Union’s Artificial Intelligence Act[18] and the Singapore Infocomm Media Development Authority (IMDA) Model AI Governance Framework. The CGH AI Governance Framework was developed with four key considerations [Box 1] and provides a comprehensive approach to ensuring trustworthy AI development and deployment using a risk-based approach [Figure 2].

Box 1.

Key considerations in the development of the AI Governance Framework

1. Regulatory compliance: Ensures adherence to the AI Act’s requirements, including transparency, explainability and human oversight

2. Ethics and values: Incorporates principles from the EU’s Ethics Guidelines[18] for trustworthy AI, such as respect for autonomy, fairness and human dignity

3. Technical standards: Leverages industry-recognised standards, such as those provided by Dataiku’s Generative AI framework, to ensure technical robustness and reliability[19]

4. Accountability and oversight: Establishes mechanisms for monitoring, reporting and addressing AI-related risks and incidents

AI: artificial intelligence, EU: European Union

Figure 2.

The AI Governance Framework.

BUILDING STRONG INFORMATION TECHNOLOGY INFRASTRUCTURE

Since the hospital’s establishment, it has undergone significant digital transformation, evolving from traditional information technology (IT) systems to a sophisticated AI-powered healthcare ecosystem. This journey has been driven by the need to enhance patient care, optimise hospital operations and ensure robust cybersecurity against emerging threats. The foundation of our digital transformation began with a structured IT roadmap, progressively integrating data analytics, cloud computing and AI capabilities into our healthcare system. The implementation of an enterprise cloud infrastructure facilitated secure data storage, AI model training and seamless interoperability between various digital health systems.

A notable project in our IT evolution is the introduction of an integrated bed management solution. By leveraging radio frequency identification technology to track patient flow, update bed statuses and trigger housekeeping upon discharge, the system has digitalised the entire inpatient flow process, enhancing both efficiency and patient experience.

Our hospital’s digital transformation journey has also been enabled by close collaboration with key partners and stakeholders within the national public healthcare ecosystem. Synapxe, the central IT provider for all PHIs in Singapore, played a critical role in co-developing and operationalising hospital-level digital solutions such as the enterprise cloud infrastructure, business and clinical systems enhancements and AI model deployment pipelines. In addition, within the SingHealth cluster, cross-institutional learning and shared infrastructure — such as participation in cluster-wide data lakes and digital governance working groups — have accelerated CGH’s adoption of smart hospital practices. We also actively engage in national efforts led by the MOH and Synapxe, contributing to initiatives such as the AI Medical Imaging platform for Singapore Healthcare (AIM.SG), National Analytics Platform, HEALIX, National Electronic Health Record and cybersecurity resilience programmes across clusters.

Our electronic medical record (EMR) system has continuously evolved to support real-time clinical decision-making. Initially developed to digitise patient records, it has been enhanced with AI-driven automation, interoperability features and integration with national health databases. The implementation of ambient digital scribing tools, such as Note Buddy (a generative AI-powered solution) and Pair (a Singapore government AI chatbot assistant), has streamlined medical documentation, reducing the administrative burden on clinicians while ensuring data accuracy and compliance with privacy regulations.[20] In addition, we are aligning with the national initiative to adopt the the Epic EMR system — developed by the international software company Epic — with the aim of unifying the entire public health system on a common platform, a pioneering move globally. We have also deployed commercial AI radiology software on AIM.SG, a data platform integrated with the hospital’s picture archiving and communication system (PACS). Data platforms are essential for integrating AI software with hospital systems such as PACS, as they lower the barriers to adoption and ensure interoperability across existing hospital information systems.

BUILDING RESILIENT CYBERSECURITY SYSTEMS

As digitalisation expanded, the need for a comprehensive cybersecurity strategy also grew. Recognising healthcare as a prime target for cyber threats, our hospital has prioritised enterprise risk management, AI governance and cybersecurity resilience. We have adopted a zero-trust security architecture, real-time threat detection and robust AI-driven anomaly detection systems. Lessons learned from global cyber incidents have underscored the importance of supply chain security, ransomware mitigation and regulatory compliance with national cybersecurity standards. Regular staff training in cybersecurity and risk management further ensures that all stakeholders understand and adhere to best practices in mitigating cyber risks.

DRIVING ARTIFICIAL INTELLIGENCE LITERACY

There is a need to equip individuals with the knowledge and skills to understand, use and interact with AI responsibly and effectively. By establishing a fundamental understanding of AI, healthcare organisations can then unlock its full potential to drive innovation, efficiency and growth. This is a structured approach to Enterprise Data and AI Literacy.[21] By investing in AI literacy programmes, healthcare organisations can bridge the AI knowledge gap, foster a culture of AI-driven innovation and drive successful AI transformation.

Our hospital offers three major programmes in AI literacy for staff catered to different levels of AI competency. At the foundation level, all staff — from executive and clerical to senior management — undergo a foundational AI literacy course conducted by AI Singapore, a national programme aimed at growing local AI talent and building an AI ecosystem. At the beginner level, the foundational course is complemented by intermediate skill acquisition programmes conducted by Duke-NUS and Singapore Institute of Technology. For practitioner-level users, the hospital partnered with Singapore University of Technology and Design to co-develop and conduct a ModularMaster certificate programme in Data Science and AI with a focus on healthcare. Having multiple partners in creating educational content allowed us to select and match the relevant modules to scale staff training in AI and tap on the strengths of the various partners. This was achieved by clearly defining the different levels of competencies.

ESTABLISHING A MULTIDISCIPLINARY TEAM

We adopted an integrated multidisciplinary model by establishing an AI and Digital (AID) committee, comprising expertise from clinical teams (medicine, nursing and allied health), data and technology teams, research and innovation offices, along with other key stakeholders [Figure 3].[22]

Figure 3.

Changi General Hospital’s AI and Digital ecosystem. AI: artificial intelligence, AID: AI and Digital, CHART: Centre for Healthcare Assistive and Robotics Technology, CIO: Chief Information Officer’s office, CT: Care Transformation, DSI: Data Science and Intelligence, HSR: Health Services Research, OOI: Office of Innovation, OOR: Office of Research, ORM: Office of Risk Management.

The role of the AID committee is to drive the development and adoption of AI and digital solutions in healthcare [Figure 4]. A hospital AID portal was also established to allow users to contribute to use cases and learn from both success and ‘failure to launch’ stories. The incorporation of the Office of Risk Management was strategic to ensure that risks (including AI enterprise risks) are mitigated using the enterprise risk management framework. Strong support from the hospital leadership and management has also been crucial, providing funding and effective staff engagement through townhalls and webinars.

Figure 4.

Role of Changi General Hospital’s AI and Digital (AID) committee. CHART: Center for Healthcare Assistive & Robotics Technology, CT: care transformation, DL: deep learning, DSI: data science & intelligence, HSR: Health Services Research, ITSC: Information Technology Steering Committee, ML: machine learning, MOH: Ministry of Health, OOI: Office of Innovation, OOR: Office Of Research, SHS: SingHealth Services.

Integration with the Office of Innovation allowed us to leverage strong local, regional, and international partnerships with hospitals, institutions of higher learning, industry and start-ups. Building on these collaborations enabled the rapid establishment of the AIDE.

BUILDING AN INNOVATIVE AND COLLABORATIVE CULTURE

Healthcare innovation is one of the key elements supporting a sustainable healthcare system, especially against the backdrop of a rapidly ageing population. In 2012, CGH was the first public hospital in Singapore to establish a Centre for Innovation (CFI), seeded by the Economic Development Board of Singapore as a super platform to support collaboration with industry and academic partners in healthcare innovation. The CGH Office of Innovation (OOI) was formed in 2015 to better coordinate and strategise the activities undertaken by CFI and other centres of excellence driving innovation. In 2015, CGH embarked on digital health, years ahead of its rapid development during the COVID-19 pandemic. A trial was conducted in partnership with industry to ascertain the effectiveness of telemonitoring for heart failure patients. The study concluded that telemonitoring was associated with lower all-cause and heart failure-related total bed days at 180 days, lower heart failure-related total bed days and total cost of care at 1 year compared to structured telephone support.[23]

The need for digital health and AI-related technologies to augment clinical care and productivity grew during and after the COVID-19 pandemic. The OOI established a framework that supports the whole innovation life cycle — from identifying unmet needs and fostering collaborations with external partners to co-create solutions, to test-bedding these solutions in real-world settings as proof of concept or value, and finally facilitating early adoption and scaling. It took the centre stage in supporting the growing needs of digital health and AI projects in the research and development phase by providing seed funding and project management expertise. At the height of the COVID-19 pandemic, CGH developed a chest X-ray (CXR)-based AI model (Community-Acquired Pneumonia AI predictive Engine [CAPE]) and prospectively evaluated its discrimination for 30-day mortality. It was proven that CXR-based CAPE mortality risk score was comparable to traditional pneumonia severity scores and improved its discrimination when combined.[24] The challenges presented by the pandemic led to the search for more effective infection control and remote patient monitoring solutions. Clinicians also collaborated with Respiree Pte Ltd (a local start-up) and the Agency of Science and Technology and Research in the validation and deployment of real-time remote vital sign monitoring wearables.[25] Algorithms have been developed to better predict the deterioration of patients in general wards for early intervention and optimisation of healthcare resources.

The OOI has played an important role as a control tower, facilitating partnerships between CGH innovators and external partners from institutions of higher learning and industry. With a mission to deliver value and transform care, CGH aims to collaborate closely with our partners to enable knowledge and technology transfer in healthcare digital innovation and AI. By fostering these collaborations, the translation of digital technologies and innovations into clinical adoption can be more effective and efficient, ultimately delivering better healthcare and economic outcomes. Figure 5 outlines the interface and synergy between the hospital’s Research, Innovation and Enterprise committee and the AID committee.

Figure 5.

Interface and synergy between the Research, Innovation and Enterprise (RIE) committee and AI and Digital (AID) committee.

ACCELERATING DATA SCIENCE AND AI THROUGH RESEARCH AND INNOVATION

Singapore public hospitals face five major challenges when embarking on AI and data science projects: (1) fragmented data across systems with quality issues; (2) strict data governance and security requirements; (3) competing hospital priorities requiring hospital staff to prioritise meeting clinical service demand over other tasks or projects; (4) prohibitive budget requirements for IT system integration or support; and (5) change management issues such as securing staff buy-in for adoption of AI and data science tools.

Despite these challenges, our hospital has been actively developing and implementing AI and data science tools, leading to a proliferation of such tools and projects, particularly in diagnostic support, operational management and prioritisation of patients for clinical care. In diagnostic applications, several studies have demonstrated AI’s effectiveness and potential in medical imaging analysis. In orthopaedics, DL algorithms[26,27] successfully classified and localised implant cutouts in postoperative hip cases as well as detected hip fractures. In gastroenterology, a validation study of a computer-aided diagnosis system[28] for predicting polyp histology during colonoscopy was performed to understand the future developmental opportunities in this area. In nursing, an AI-driven wound care management solution was implemented hospital-wide (emergency department [ED], inpatient wards, outpatient clinics and community nursing team) to enable nurses to adopt a more efficient and productive wound management process. The wound care system provides dynamic visualisation of the wound condition and healing process using graphs, visible images and alerts.

In the realm of operational management, an ML model[29] was developed to predict short inpatient stays among urgent admissions, which can potentially help with hospital resource planning. An automated bed assignment algorithm[30] was also implemented in a tertiary hospital in Singapore, enabling a more consistent and timely assignment of beds, so that patients can receive optimal inpatient care at the wards. Another study[31] identified risk factors for hospital admissions from ED attendances, ensuring better prediction of inpatient admission needs that originate from the ED.

On the manpower allocation front, CGH ED has implemented a junior doctor manpower allocation plan derived through optimisation modelling.[32] Without increasing overall manpower requirements, this initiative enabled the ED to reduce time to first consult, thereby improving the patient experience.

The CGH team has also examined factors predicting return to work following inpatient stroke rehabilitation, using data analysis[33] to identify patients who might need additional support for successful rehabilitation outcomes. Figure 6 shows a summary of data science and AI success stories [see Supplemental Digital Appendix for more details].

Figure 6.

Summary of Data Science and AI success stories in Changi General Hospital. CAPE: Community Acquired Pneumonia and COVID-19 AI Predictive Engine, CXR: chest X-Ray, DL: deep learning, ED: emergency department, EDW: enterprise data warehouse, eHINTS: Electronic Intelligence System, GP: general practitioner, iHAP: Integrated Health Analytics Platform, IP: inpatient, ISO: International Standards Organization, MET: medical emergency team, ML: machine learning, OT: operation theatre, RPA: robotic process automation, SOC: specialist outpatient clinics, TCF: transition care facility.

ORGANISATIONAL LEADERSHIP

Key leadership strategies that contributed to the success of the data-driven journey include; (1) strengthening leadership knowledge in data science and AI at all levels; (2) nurturing a culture of continuous learning, innovation and collaboration; (3) fostering multidisciplinary collaboration among hospital staff, as well as with industry and other stakeholders; (4) focusing not only on development of data and AI solutions but also their integration and adoption; and (5) building both leadership capacity and a data science and AI talent pipeline.

At the leadership level, several challenges remain: (1) keeping pace with rapid advancements in data science and AI; (2) assessing the cost-effectiveness and return of investment of digital solutions, which are not yet clearly defined within the healthcare system; (3) managing the cost of integrating digital solutions within the hospital’s secured IT system; and (4) driving effective change management. To date, our hospital has conducted a simple cost-avoidance analysis following the deployment of a chest radiograph triaging and reporting AI software, to assess its impact on report generation times and radiologist manpower hours. These challenges are complex, and potential solutions will take time to design and implement.

OPPORTUNITIES AND CHALLENGES

Despite the potential applications of AI in healthcare, several ‘unknown unknowns’ remain. The US Food and Drug Administration is developing an AI governance, risk management and ethical framework and an approval process for large language models used in clinical practice. Our hospital has been collaborating with national agencies, such as the Health Sciences Authority and IMDA, to validate AI models deployed in healthcare. Changi General Hospital is the first PHI to join AI Verify Foundation to build and validate trustworthy AI solutions. These AI models can then be shared, scaled and adopted across other local hospitals and beyond. Hospitals can also adopt best practices and AI standards such as ISO 42001 (management system), ISO 23894 (risk management of AI), ISO 25336 (conformity assessment schemes for AI) and ISO 24029 (robustness of neural networks).[34] Indeed, there are more opportunities than challenges. Ten key recommendations for successful establishment of an AIDE are presented in Box 2.

Box 2.

Ten key recommendations for successful establishment of an AIDE.

1. Build a strong foundation in data management, informatics and data science

2. Ensure strong data governance and cybersecurity systems

3. Develop the talent pipeline of data analysts and data scientists

4. Establish a transdisciplinary team

5. Build strong partnerships and collaboration with industry, IHLs and other stakeholders as part of a larger AIDE

6. Obtain strong support from leadership and management, including infrastructure and funding support, to catalyse adoption or implementation

7. Drive AI and digital literacy in your healthcare organisation

8. Focus on the adoption of AI and digital solutions, not just the development

9. Adopt AI and digital solutions that are patient-focused, cost-effective and cyber-safe

10. Adopt a robust AI governance framework and ERM system

AI: artificial intelligence, AIDE: AI and Digital Ecosystem, ERM: enterprise risk management, IHLs: institutions of higher learning

FUTURE OF ARTIFICIAL INTELLIGENCE

Artificial intelligence is set to revolutionise medicine, from diagnostics to personalised treatment. Integrating AI into healthcare will improve efficiency, accuracy and patient outcomes. Changi General Hospital is contributing to the development of several major areas that will emerge as growth drivers in Singapore’s biomedical science sector, powered by AI. We highlight below some areas that will likely transform our healthcare practices in the future.

Workforce education and metacognitive training

Integrating AI in healthcare requires a workforce skilled in AI technologies. The National Academy of Medicine highlights the need to promote an AI-competent healthcare workforce, including the incorporation of AI curricula in medical education.[35] Metacognitive training, which involves teaching clinicians to think about their own thinking processes, is crucial for the effective use of AI. This training helps clinicians to understand AI’s decision-making processes, thereby improving their ability to collaborate with the AI systems and make informed clinical decisions.[36,37] At CGH, we have designed a framework for clinicians to understand cognitive biases and to apply metacognitive thinking when using AI.

Pandemic preparedness

Artificial intelligence can greatly improve pandemic preparedness by enhancing disease surveillance, outbreak prediction and response strategies. Algorithms analyse data from various sources, including social media, health records and data registries, to detect early outbreak signs and forecast spread, enabling timely interventions and better resource allocation to reduce impact.[38] At CGH, we have developed an ensemble AI model that can automatically analyse CXRs to predict the probability of intensive care unit admission of patients with COVID-19, enabling early intervention of the sickest patients. We have also co-developed a 4D digital twin model of the built environment at inpatient wards with Singapore General Hospital to predict the spread of pathogens and enhance contact tracing within the hospital.

National AI medical imaging platform

Robust IT infrastructure and reliable networking capabilities form the essential foundation for developing, training and deploying AI systems effectively, while ensuring they remain secure, accessible and robust without model decay. In 2019, the blueprint for a national IT platform (AIM.SG) was developed to enable public hospitals to access different imaging AI models through one single platform and to monitor their performance. This platform was operational in August 2023 at CGH and is now deployed across SingHealth.[39] This has enabled our hospital to test and validate CXR AI models using real-world information.[40]

Agentic AI and AI agents

Agentic AI in healthcare refers to the use of AI systems that can autonomously perform tasks, make decisions and adapt to new situations, similar to a human agent in clinical and non-clinical settings. These agentic AI systems are designed for autonomy, agency, intentionality and flexibility. Applications in healthcare are diverse and not limited to clinical decision support, personalised medicine, predictive analytics, robot-assisted surgery and patient engagement. However, the deployment of AI agents and agentic AI workflows present challenges and limitations. These include the need for compliance with regulatory frameworks to ensure safety, efficacy and accountability; data quality and integration; explainability and transparency; and cybersecurity.[41] With the surge in open-source AI agents for specific tasks, a cautious, risk-based approach should be adopted to pilot AI agents in low-risk settings with a focus on administrative backend tasks. Additionally, a hybrid human-in-the-loop approach will be needed in moderate- to high-risk healthcare settings.[42,43]

CONCLUSION

The call to action is now. The AI and digital landscape will continue to evolve rapidly. As hospitals progress in developing and adopting AI and digital solutions, the pace of transformation and change management will differ across healthcare institutions. Strong leadership and management support, a robust ecosystem, and most importantly, a workforce skilled in data science and AI are critical success factors for driving a data-driven, thriving healthcare system towards excellence.[44,45]

Conflicts of interest

There are no conflicts of interest.

Supplemental digital content

Appendix at http://links.lww.com/SGMJ/A236

APPENDIX

SUMMARY TABLE OF DATA SCIENCE AND AI SUCCESS STORIES

	Domain	Name	Year	Speciality Field	Description	Impact	Publications/ Recognition/ LIVE Deployment
1	AI/ML/DL	Prediction of Short IP stay	2025	Inpatient bed management	Support risk stratification of ED patients who require inpatient admissions	Allows hospital to identify inpatient short-stayers who may potentially be targeted for new programmes which can alleviate inpatient bed demand via community-based interventions	Gao Y, Srivatsava S, Oh HC, Goh SH, Lim SHC. Development and Validation of a Machine Learning Model That Predicts Short Inpatient Stays Among Urgent Admissions. Emerg. Care Med. 2025, 2(1), 11; https://doi.org/10.3390/ecm2010011
2	AI/ML/DL	Generative AI	2024	Generative AI	Summarization of Colonoscopy and Histology report for Gastroenterology specialists	Productivity savings and accuracy of clinical report summarisation for care excellence and decision support	Pending, AI Assurance Pilot evaluation in progress
3	AI/ML/DL	CXR Triage AI Model (LUNIT)	2024	Medical Imaging	AI Impact Assessment: CXR triage AI model validated in a real-world clinical prospective study	It was demonstrated that AI triaging of CXR was accurate, and able to reduce turnaround time of radiology reports. Deployed in clinical radiology IT system (PACS) on AimSG national AI imaging platform	Sridharan S, Hui AS, Venkataraman N, Tirukonda PS, Jeyaratnam RP, John S, Babu SS, Liew P, Francis J, Tzan TK, Min WK, Liew CJ. Real-World evaluation of an AI triaging system for chest X-rays: A prospective clinical study. European Journal of Radiology. 2024 Oct 10.
4	AI/ML/DL	Hip Fracture detection	2023	Medical Imaging	Development of AI model for detection of hip fracture in x-ray images	Potential more timely diagnosis of hip fracture in ED setting	Tan JR, Gao Y, Raghuraman R, Ting D, Wong KM, Cheng LT-E, Oh HC, Goh SH, Yan YY (2024) Application of deep learning algorithms in classification and localization of implant cutout for the postoperative hip. Skeletal Radiol. https://doi.org/10.1007/s00256-024-04692-6
5	AI/ML/DL	Hip Fracture detection v2	2023	Medical Imaging	Development of AI model for detection of hip fracture in x-ray images	Potential more timely diagnosis of hip fracture in ED setting	Gao Y, Soh NY, Liu N, Lim G, Ting D, Cheng LT, Wong KM, Liew C, Oh HC, Tan JR, Venkataraman N. Application of a deep learning algorithm in the detection of hip fractures. Iscience. 2023 Aug 18;26(8).
6	AI/ML/DL	Procedure - Colonoscopy	2022	Procedure - Colonoscopy	Validation of decision support tool to diagnose polyp histology in colonoscopy in a real-world setting	Current decision support tool has the potential to be enhanced further so that the diagnostic accuracy and sensitivity for neoplastic polyps are comparable with those of experienced endoscopists.	Li JW, Wu CC, Lee JW, Liang R, Soon GS, Wang LM, Koh XH, Koh CJ, Da Chew W, Lin KW, Thian MY. Real-World Validation of a Computer-Aided Diagnosis System for Prediction of Polyp Histology in Colonoscopy: A Prospective Multicentre Study. The American Journal of Gastroenterology. 2022 May 12:10-4309.
7	AI/ML/DL	CAPE	2021	AI Imaging	Chest X-Ray (CXR)-based AI model (Community-Acquired Pneumonia AI predictive Engine: CAPE) and prospectively evaluated its discrimination for 30-day mortality.	It was proven that CXR-based CAPE mortality risk score was comparable to traditional pneumonia severity scores and improved its discrimination when combined. CAPE was validated on the AimSG national AI imaging platform	Chest radiograph-based artificial intelligence predictive model for mortality in community-acquired pneumonia BMJ Open Respir Res. 2021 Aug;8(1):e001045. doi: 10.1136/bmjresp-2021-001045.
8	AI/ML/DL	Respiree	2021	Med Tech	Respiree - Medical Device Development and Machine learning model for predicting severity and acuity escalation for MET/Code Blue scenarios	Patient safety alert systems	Pending publication, SG President’s Science and Technology Award
9	Data Science	Inpatient bed demand from ED patients	2024	Inpatient bed demand from ED patients	Identify risk factors for hospital admissions among emergency department attendees	Findings from such analysis offer invaluable insights on how new programmes can be designed and developed to mitigate future inpatient bed demand among ED patients	Koh J, Oh HC, Seah Z, Lim S. Risk factors for hospital admissions among emergency department attendees: From triage to admission. (Accepted by Western Journal of Emergency Medicine in Dec 2024).
10	Data Science	Healthier SG Program evaluation	2024	Healthier SG	Impact evaluation of Preventive programs under Healthier SG	Cross institutional validation of Healthier SG outcomes	Yong Yang , Indumathi Venkatachalam, Chi Ting Low, Mabel Zhi Qi Foo, May Kyawt Aung, Shawn Wee Jin See, Myat Oo Aung, Darius Yak Weng Chan, Shalvi Arora, Jean Xiang Ying Sim, Yuke Tien Fong , Vui Kian Ho, Yee Sien Ng, Lian Leng Low, Srinath Sridharan , Moi Lin Ling , Chai Rick Soh, Transforming healthcare system: Outcomes of Healthier-SG from a large tertiary-care hospital in Singapore,Health Policy and Technology 14 (2025) 100968
11	Data Science	Inpatient bed management	2023	Inpatient bed management	Development of a decision support tool which can automate the bed assignment decisions which were otherwise performed manually of Bed Management Unit staff	Patients are assigned to inpatient beds in a more consistent and timely manner so that they receive optimal inpatient care at the wards	Oh HC, Ng SC, Zeng ZZ. Developing and evaluating an automated bed assignment algorithm in a tertiary hospital: A case study in Singapore. Asia Pacific Journal of Health Management. 2023 Jun 1;18(2):148-63.
12	Data Science	Inpatient stroke rehabilitation	2022	Inpatient stroke rehabilitation	Identify factors predicting return to work following inpatient stroke rehabilitation	Findings from such analysis offer invaluable insights on how new programmes can be designed and developed to facilitate return to work among patients following inpatient stroke rehabilitation	Tay SS, Visperas CA, Tan MM, Chew TL, Koh XH. Factors Predicting Return to Work Following Inpatient Stroke Rehabilitation: A Retrospective Follow-Up Study. Archives of Rehabilitation Research and Clinical Translation. 2022 Dec 24:100253. https://doi.Org/10.1016/j.arrct.2022.100253
13	Data Science	ED manpower allocation	2017	ED manpower allocation	Employed optimization modelling to determine ED doctor manpower allocation plan which could reduce time to consult among ED attendees	Improved ED patient experience via reduction of time to consult intervals of ED attendees	Oh HC, Chow WL, Looi, P, Goh PL, Lim SHC, Tiruchittampalam M. Patient Experience Enhancement at Accident and Emergency Department: Junior Doctor Manpower Reallocation Optimization. Journal of Management Science and Engineering 2017; 2(3), 193-208.
14	Informatics, Analytics & Automation	Sustainability innovation in OT for Nitrous Oxide	2024	Operations Theatre	Reduction of Nitrous oxide wastage translating to environmental sustainability	Sustainability in healthcare - operations theatre perspective	Michelle Bee Hua Tan, William Gagnon , Huae Min Tham, Li Wen Ong,Bernard Tiang Guan Koh , Srinath Sridharan , Xuan Han Koh and Jo-Anne Yeo, Utilising electronic anaesthesia records to audit nitrous oxide consumption and waste, and implementation of a loss monitoring system, British Journal of Anaesthesia, https://doi.org/10.1016/j.bja.2024.12.018
15	Informatics, Analytics & Automation	Roster Optimisation	2024	Productivity	Optimization of Roster for Cardiology and Pharmacy with RPA automation for productivity gains	Productivity savings and decision support	Live solution Implemented, Pending publication
16	Informatics, Analytics & Automation	RPA for Financial Counselling	2024	Productivity	RPA for Inpatient Operations team for e- financial counselling for patients	Productivity savings for Inpatient Ops team	Live solution Implemented, Poster published
17	Informatics, Analytics & Automation	RPA for Pharmacy	2024	Productivity	RPA for Pharmacy Tech workflow automation for medication orders from Inpatient settings	Productivity savings for Pharmacy team	Live solution Implemented, Pending publication
18	Informatics, Analytics & Automation	RPA for Schimadzu Lab C3	2024	Productivity	RPA for Lab technician process workflow	Productivity savings for Clinomics Centre team	Live solution Implemented, Pending publication
19	Informatics, Analytics & Automation	TCF Dashboard	2024	TCF Dashboard	Tracks transitional care facility metrics, including patient transfers, length of stay, and bed availability	TCF Dashboard has revolutionised transitional care management at the hospital by optimising bed utilisation and streamlining patient transfers. The system has significantly reduced acute care length of stay while improving care continuity for transitional patients, resulting in more efficient use of healthcare resources and enhanced patient care pathways.	Live solution Implemented
20	Informatics, Analytics & Automation	4D-Disease Outbreak Surveillance System (4D- DOSS)	2021/ 2023/ 2024	Digital Twin	Development and implementation of digital twin hospital system for spatiotemporal mapping for infectious disease surveillance and outbreak investigations including assessment of cost -effectives of the digital twin surveillance system	The integration of health care data and representation on a virtual hospital digital twin is a useful tool in an outbreak alert and response framework. Infectious disease surveillance systems, which are syndrome-based, can access real-time data, and can incorporate movement networks, can potentially enhance health care associated infection prevention and preparedness for disease X. 4D-DOSS was successfully implemented in another hospital and plan for extension to other public healthcare institutions.	Indumathi Venkatachalam, Edwin Philip Conceicao, Jean Xiang Ying Sim, Sean Douglas Whiteley, Esther Xing Wei Lee, Hui San Lim, Joseph Kin Meng Cheong, BE; Shalvi Arora, Andrew Hao Sen Fang, and Weien Chow. Three-Dimensional Disease Outbreak Surveillance System in a Tertiary Hospital in Singapore: A Proof of Concept. Mayo Clinic Proceedings: Digital Health June 2023 1(2):172-184 Cai Y, Philip EC, Arora S, Sim JXY, Chow W, Nazeha N, Whiteley S, Tiang DC, Neo SL, Hong W, Venkatachalam I, Graves N. The attributable mortality, length of stay, and health care costs of methicillin-resistant Staphylococcus aureus infections in Singapore. IJID Reg. 2024 Aug 21;12:100427. doi: 10.1016/j.ijregi.2024.100427. PMID: 39281193; PMCID: PMC11402250.
21	Informatics, Analytics & Automation	MET Code Blue Dashboard	2023	MET Code Blue Dashboard	Monitors and tracks medical emergency team responses and code blue incidents throughout the hospital.	The MET Code Blue Dashboard has shed visibility on hospital’ emergency response capabilities by enabling faster and more coordinated medical emergency team deployments. This system has enhanced the tracking of emergency outcomes, improved team coordination and provided valuable insights for continuous improvement of emergency response protocols.	Live solution Implemented
22	Informatics, Analytics & Automation	Nursing Home Dashboard	2023	Nursing Home Dashboard	Interface for monitoring hospital’s partnerships with nursing homes, including patient transfers and care coordination	Review of Nursing Home care performance for Case Management team and improving readmission rates for NH cohort	Live solution Implemented
23	Informatics, Analytics & Automation	COVID Contact Tracing	2020	COVID Contact Tracing	A contact tracing system implemented to track and manage potential COVID-19 exposures within the hospital, helping identify and notify staff and patients who may have been in contact with confirmed cases.	COVID Contact Tracing has revolutionised hospital's pandemic response capabilities, establishing a robust system that significantly reduced in-hospital COVID-19 transmission rates. The implementation of this digital contact tracing infrastructure not only protected vulnerable patients and healthcare workers but also enabled rapid isolation of potential cases, maintaining the hospital's operational resilience during critical periods of the pandemic.	Venkataraman N, Poon BH, Siau C. Innovative use of health informatics to augment contact tracing during the COVID-19 pandemic in an acute hospital. J Am Med Inform Assoc. 2020 Dec 9;27(12):1964-1967. doi: 10.1093/jamia/ocaa184. PMID: 32835358; PMCID: PMC7499570.
24	Informatics, Analytics & Automation	COVID ISO Dashboard	2020	Covid ISO Dashboard	A monitoring dashboard showing isolation ward capacity, patient status, and resource allocation for COVID-19 cases.	The COVID ISO Dashboard transformed hospital’s management of isolation facilities, providing real-time visibility of isolation bed capacity and resource utilisation. This innovative solution enabled hospital leadership to make data-driven decisions during pandemic peaks, resulting in optimised patient care and more efficient allocation of critical resources, ultimately enhancing the hospital’s ability to manage COVID-19 cases effectively.	Live solution Implemented
25	Informatics, Analytics & Automation	Smartview SOC Dashboard	2020	Smartview SOC Dashboard	A weekly monitoring system for the hospital’s Specialist outpatient Clincs, providing visibility of patient volume at clinics to allocate resources and manage patients expections.	The Smartview OT Dashboard has significantly enhanced surgical operations at the hospital, driving a marked improvement in operating theatre utilisation rates. By providing visibility of surgical schedules and resource availability, the system has reduced cancellations and delays, optimised surgical resource allocation and improved overall theatre efficiency.	Live solution Implemented
26	Informatics, Analytics & Automation	Smartview OT Dashboard	2020	Smartview OT Dashboard	Tracks operating theatre utilisation, scheduling and real-time status of surgical procedures	Optimisation of OT usage during Pandemic, and review of phased return to BAU status	Live solution Implemented
27	Informatics, Analytics & Automation	Marketshare Analysis	2020	Marketshare Analysis	Analytics tool for understanding the hospital’s market position and patient demographics in the eastern region	Market share Analysis has provided the hospital with crucial insights into community healthcare needs and patient demographics. This analytical tool has enabled more targeted service development and enhanced strategic planning, allowing the hospital to better serve its catchment area and adapt its services to meet evolving healthcare demands.	Live solution Implemented, for Annual Market Review and Sr Mgmt Decision Support
28	Informatics, Analytics & Automation	ED Live Dashboard	2014	ED Live Dashboard	Real-time monitoring of Emergency Department patient flow, waiting times, bed occupancy and resource allocation	The ED Live Dashboard has transformed emergency care delivery at the hospital by providing real-time monitoring of patient flow and resource utilisation. This system has significantly reduced waiting times, improved patient flow management and enhanced communication between emergency and inpatient teams, resulting in more efficient emergency care delivery and better patient outcomes.	Live solution Implemented, National Elsevier Award
29	Informatics, Analytics & Automation	GP First Program	2014	GP First Program	Initiative to strengthen partnerships with General Practitioners in the region, including referral management and care coordination	The GP First Program has strengthened hospital’s primary care network by fostering closer partnerships with community doctors. This initiative has successfully reduced unnecessary ED visits, improved care coordination and enhanced continuity of care, demonstrating the hospital’s commitment to integrated healthcare delivery in the eastern region of Singapore.	Oh HC, Sridharan S, Yap MF, Goh PSK, Lee LSH, Venkataraman N, How CH, Lim HC. Impact of a primary care partnership programme on accident and emergency attendances at a regional hospital in Singapore: a pilot study. Singapore Med J. 2023 Aug;64(8):534-537. doi: 10.11622/smedj.2021157. PMID: 34628785; PMCID: PMC10476921.
30	Data Management	EDW- eHINTS	2016	Enterprise Data warehousing - eHINTS	Harmonisations of EDW with SHS eHints	Support stat submission for MOH SHS and RIE projects, Disease Registries.	Live solution Implemented
31	Data Management	Hospital Clearinghouse and Data Governance Framework	2012	Hospital Clearinghouse Framework	Central Data Clearinghouse function for streamlining data mining, extraction and processing to ensure robust data governance and security	Reduction in IT Opex for IHIS/Synapxe towards data mining, faster processing times for hosptial data requests and RIE projects, deepening hospital’s data capabilities.	Live solution Implemented
32	Data Management	EDW - iHAP	2011	Enterprise Data warehousing - iHAP	Enhancement of EDW for Oracle and Informatica	Support stat submission for MOH SHS and EHA clusters	Live solution Implemented
33	Data Management	Hospital Workload Statistics Summary Report	2010	Workload statistics	Summary of Hospital Monthly Workload statistics	Comprehensive summary of workload across all care domains with speciality level insights	Live solution Implemented
34	Data Management	EDW - Sagent	2008	Enterprise Data warehousing - Sagent	Central Data warehouse for Hospital	Support stat submission for MOH SHS and EHA clusters	Live solution Implemented

Funding Statement

Nil.