Topic 1: Using Artificial Intelligence to Address Imminent Challenges in Health Care
Project Title: Personalized Rehabilitation Pathways to Maximal Motor Functional Return through an AI Recovery Prediction System for Diverse Stroke Survivors
Project Coordinator: Prof Vincent Chi-kwan Cheung (CUHK)
Abstract
This project addresses the imminent challenge of providing adequate motor rehabilitation to a growing number of stroke survivors amidst the ageing population, younger onset of stroke, and shortage of physical/occupational therapists in Hong Kong through AI and precision rehabilitation.
Stroke is the leading cause of adult disability in Hong Kong and the third globally. To reduce the socioeconomic burden from the stroke survivors’ dependence and care needed (est. >HK$15 billion/year), the efficacy of rehabilitation and efficiency of its delivery must be improved. These goals can be achieved by prescribing them with tailored rehabilitations predicted to yield maximal functional return. Defining a predictive model for such personalization remains challenging given the immense heterogeneity of stroke patients. We aim to build a preliminary version of an explainable AI system, called PRAISE-HK (PoC) (Precision Rehabilitation AI System for Enhanced recovery in Hong Kong and beyond, Proof of Concept version), that predicts a subject’s recovery potential and the treatment option that has the maximal likelihood to realize this potential based on multi-modal pre-rehabilitation assessments.
Clinical, neuroimaging, neurophysiological, and multi-omics data will first be collected from subacute stroke survivors (N=200-400) before they undergo upper limb rehabilitation with usual care, acupuncture, robotic training, or neuromuscular stimulation (Deliverable 1). Not only are these interventions relatively accessible in Hong Kong, but tremendous resources have also been previously devoted to their development. Machine learning-extracted features from these data will then be used to train neural-network and decision-tree algorithms in PRAISE-HK (PoC) for robust predictions of post-intervention motor recovery levels for different treatments (Deliverable 2). After validating the system, we will deploy it to implement a pilot personalized rehabilitation program in the community (Deliverable 3).
Our model’s potential ability to predict the optimal intervention, out of multiple treatment options, from a wide spectrum of input modalities will distinguish ours from previous less-than-accurate models. Our exceptionally interdisciplinary team of 13 PC/Co-PIs with expertise in neurology, PT/OT, acupuncture, electrical/biomedical engineering, robotics, neuroscience, neuroimaging, multi-omics, data science, and clinical trial management will put us in a world-unique position to execute this project and generate opportunities of interdisciplinary education. In the long run, the eventual full version of PRAISE-HK will accelerate marketization of new rehabilitative strategies by facilitating their clinical-trial evaluations in more targeted populations. Such ease of conducting rehabilitation trials and the large scope of this project will attract local and international, industrial and academic rehabilitation scientists to cross-fertilize and develop new rehabilitative devices in Hong Kong, leading Hong Kong to be a future global hub of innovative rehabilitation.
Topic 1: Using Artificial Intelligence to Address Imminent Challenges in Health Care
Project Title: Foundation Models-Empowered Ambient Intelligence Systems for Early Diagnosis, Personalized Intervention, and Complex Cross-Disease Interplay Analysis of Aging-Related Degenerative Diseases
Project Coordinator: Prof Guoliang Xing (CUHK)
Abstract
Aging-related degenerative diseases including dementia and sarcopenia have become a global health challenge due to prominent population aging. Each condition alone is associated with an increased risk of mobility disorders, cognitive impairment, institutionalization, and death. Moreover, recent evidence has reported the complex interplay between these two age-related conditions. This project proposes Koala, the first ambient intelligence system that leverages Foundation Models (FM), the first general AI technology, to provide non-invasive, personalized diagnosis, intervention, and cross-disease interplay analysis for aging-related degenerative diseases, including both dementia and sarcopenia. First, we will develop a multi-modal ambient sensor system deployed in the elderly home to provide longitudinal health activity detection, digital biomarker monitoring as well as natural interaction using voice and gestures with the elderly and caregivers, through effective cooperation with cloud-based FM. Second, we will design KoalaFM – the first trustworthy and data/compute-efficient FM for dementia and sarcopenia. Consisting of both a centralized FM and personalized FMs finetuned for each patient with private data, KoalaFM is capable of cross-modality knowledge fusion of highly heterogeneous digital biomarkers, medical reports, and consultation logs. Third, leveraging KoalaFM, we will propose novel approaches for the discovery of new digital biomarkers for both dementia and sarcopenia, early diagnosis and personalized intervention plans, and trustworthy medical consultation systems for doctors to enhance clinical assessment for elders with cognitive impairments during patient interactions. Koala will be validated through a large-scale comprehensive clinical trial that consists of deployments of multi-modal KoalaPortal systems as well as wearable devices. 1,000 subjects with various degrees of mobility and cognition impairments will be recruited from the out-patient clinic of three regional hospitals in Hong Kong, making it the largest cross-disease cohort study for both dementia and sarcopenia. This study will not only lead to holistic health solutions ranging from consultation, early diagnosis to intervention, but also have important implications to elucidate the synergistic evolvement of the pathological pathways between sarcopenia and dementia in a personalized context, and thereby enhance the diagnostic and therapeutic pursuits for both conditions.
Topic 1: Using Artificial Intelligence to Address Imminent Challenges in Health Care
Project Title: Development and Deployment of a Community-based Eye Care Model for Provision of Primary Eye Care Services
Project Coordinator: Prof Christopher Kai-shun Leung (HKU)
Abstract
Providing eye care services to a population of over 7 million, the Hong Kong Hospital Authority (HA) has been increasingly challenged by the aging population and manpower shortages. The shortage of ophthalmologists in Hong Kong has led to a cascade of consequences that affect patient care and broader societal productivity. While common causes of irreversible blindness like glaucoma and wet age-related macular degeneration require timely diagnosis and intervention to prevent progressive loss in vision, the longest wait time for new patient appointments in ophthalmology at the HA specialist outpatient clinics (SOPCs) is currently up to 204 weeks (90th percentile). Consequently, many patients with chronic eye diseases are undetected and untreated, ending up with permanent visual disability. There is a pressing need to revamp the model of eye care delivery via developing, deploying, and integrating a community-based system with the SOPCs, following the Hong Kong Health Bureau’s Primary Healthcare Blueprint, to improve the efficiency for the management of chronic eye diseases. We propose to develop and deploy a community-based eye care model at the District Health Centres (DHCs) – community health centres for providing primary healthcare services in various Hong Kong districts – that integrates with the SOPCs for detecting and treating chronic eye diseases. We will engineer and validate a novel artificial intelligence-augmented optical coherence tomography diagnostic system for detecting blinding eye diseases at two District Health Centres – the Southern District Health Centre and the Kwun Tong District Health Centre Express. These DHCs will be managed by ophthalmic nurses and optometrists, who will provide consultations and eye health education, manage refractive errors and dry eye disease, and make referrals for patients in need of treatment by ophthalmologists to the SOPCs. The impact and cost-effectiveness of the community-based eye care model will be investigated in a real-world pragmatic randomized clinical trial. We hypothesize that the implementation of the community-based eye care model at the DHCs will significantly reduce the time for diagnosing blinding eye diseases (primary outcome measure), with diagnostic performance that is better or on par with the current model of care. We anticipate that the community-based eye care model will substantially reduce the wait time for new patient appointments at SOPCs from up to 204 weeks to within 8 weeks. This will augment the cost-effectiveness of eye care delivery, alleviating the growing burden of visual impairment and blindness, and ultimately enhance societal productivity in Hong Kong.
Topic 2: Striving towards Carbon Neutrality before 2050
Project Title: Cost-Effective Decarbonization for the Power Sector of Hong Kong: Technology Innovation, Demonstrations, and Pathways
Project Coordinator: Prof Peng Gong (HKU)
Abstract
Achieving carbon neutrality requires comprehensive data, innovative technologies, effective economic and social policies, and active actions of stakeholders. Power generation accounts for more than 66% of the total emissions in Hong Kong (HK), demanding the biggest reduction to meet 2050 carbon neutrality goals. However, Hong Kong has yet to roll out such a concrete implementation plan due to the lack of accurate near real-time emission data, assessment of future innovative technology options and a system approach to cost-benefit analysis. Hong Kong’s carbon neutrality pathways should be assessed against the following crucial criteria: (1) The technical feasibility; (2) The power sector and energy infrastructure’s reliability and resilience; (3) Political and economic acceptability. To be cost-effective, it is essential to conduct a comprehensive evaluation of the current status and explore key future solutions for emission reduction.
We aim to achieve the following objectives:
1. Supply side: to develop high resolution top-down and bottom-up carbon emission models to identify CO2 emission sources of the power sector; based on this, to develop innovative technology portfolio for decarbonizing the power sector through the integration of renewable energy and hydrogen for carbon neutrality in Hong Kong
2. Demand side: to develop decentralized energy systems and power infrastructure planning and to comprehensively optimize and consider decentralized energy and facilities to reduce the demand for expanding power plant capacity on the demand side within the timeframe of 2035 to 2050.
3. Develop an Interactive Science-Informed Decision Model (ISIDM) for cost-effective power system planning that can be used to simulate different scenarios (e.g., technologies, decarbonization policies, and socio-ecological feedback) in search for optimal decarbonization pathways and action roadmaps for the power sector in Hong Kong, as well as similar cities in the world.
Our team is comprised of interdisciplinary experts from the University of Hong Kong (HKU), the Hong Kong Polytechnic University (PolyU), City University of Hong Kong (CityU), and the Chinese University of Hong Kong (CUHK). They specialize in areas ranging from climate change and environmental remote sensing science, Earth system science, mechanical engineering, materials science, energy engineering, power system planning, management science and engineering, carbon finance, and public policy. We are collaborating with major energy and utility/enterprise institutions in Hong Kong and the Greater Bay Area to provide cost-effective, efficient solutions for the development, assessment, and implementation of decarbonization technologies for power systems in Hong Kong and other global cities/regions characterized by “high population density + land resource scarcity.” This research is dedicated to fostering transformative innovation in renewable energy technologies within the power sector, transitioning from laboratory-scale prototypes to practical decarbonization practices, and promoting collaborative efforts in sustainability science and net-zero power systems. Our ultimate goal is to strengthen Hong Kong’s standing as an international hub for carbon neutrality and sustainable development.
Topic 3: Establishing Hong Kong as the Leading Integrated Circuits, and Opto-electronics Innovation and Technology Hub in the Guangdong-Hong Kong-Macao Greater Bay Area
Project Title: Silicon Photonics and the Heterogeneous Epitaxy of III-V Semiconductors on Silicon for Advanced Photonic Systems-on-chip
Project Coordinator: Prof Hon-ki Tsang (CUHK)
Abstract
With the imminent end to Moore’s law because of increasing technical challenges and the extreme increases in costs in manufacturing yet smaller transistors in technology nodes approaching a few atomic layers in size, we propose an alternative pathway to advance the performance of integrated systems-on-chips (SoCs) based on the use of photonic integration on silicon. We seek to develop the core technologies for advanced Photonic Systems-on-Chips (PSoCs). By processing photons instead of electrons, the photonic integration offers advantages for applications in optical sensing, data communications, high-speed signal processing, and information processing. We choose silicon as the platform of choice for the photonic integration because it enjoys the low-cost, high-yield advantages associated with CMOS manufacturing, and because of silicon’s established competitive advantages in having high-reliability and well-established supply chains, making it the dominant semiconductor for integrated circuits. In this project, we start with the research towards advancing a new process for the heteroepitaxial integration of III-V semiconductors on silicon. Our approach has the advantage, when compared with traditional blanket heteroepitaxy, of not requiring the thick buffer layers that make it problematic for integration with silicon waveguides. The proposed approach makes use of the technique of Lateral Aspect Ratio Trapping (LART) which was invented by co-PI Prof Kei May Lau. LART ensures that the III-V active region is at the same level as the silicon waveguides, enabling high-yield and low-loss coupling of light between the III-V devices and silicon waveguides. In this project we will develop the basic processes for device integration, and demonstrate the advantages of the PSoCs in enabling highly-compact systems for advanced imaging systems, energy-efficient low-latency photonic signal processors, and integrated light sources for sensing and communications. The project will use the newly developed LART in the fabrication of new monolithically integrated III-V photonic devices on silicon photonics, including the fabrication of new electrically driven lasers and broadband light emitting diodes and optical frequency comb sources on silicon photonics, III-V on silicon high-speed and energy-efficient optical modulators for data communications, III-V on semiconductor optical amplifiers and high-performance III-V photodetectors. We aim to integrate the III-V devices with high-performance passive silicon photonic functional elements such as low loss grating couplers, optical phased arrays, and integrated optical filters to build PSoCs which can be used in high-capacity optical interconnects, high-speed optical coherence tomography imaging systems, high-speed photonic accelerators for neuromorphic computing and multimode fiber imaging systems and other emerging applications.
Topic 5: Innovative and Environmental-friendly Construction Technologies and Materials
Project Title: Tomorrow is Now: Eco-friendly Autonomous Construction through Sustainable 3D Concrete Printing
Project Coordinator: Prof Kim Meow Liew (CityU)
Abstract
In Hong Kong, there is an ongoing need for construction to address the local housing crisis, as well as a lack of land for landfill development. Imminent shortages in resources and skilled labor compound these issues. Therefore, this proposed Strategic Topic project aims to develop an innovative system to construct buildings autonomously and sustainably through three-dimensional concrete printing (3DCP) with minimal interaction from a non-specialist operator. The system will reduce labor needs for the construction sector, while improving quality assurance and control, minimizing waste of materials and time, and providing a route for large-scale industrial recycling. Additionally, 3DCP excels in generating intricate shapes and structures that are unattainable through conventional construction techniques, increasing capabilities for on-site and off-site construction. This collaborative project will integrate team knowledge from multiple disciplines (materials science, computational modeling, artificial intelligence, system design, automation and control, digital twins, environmental science and public policy) to develop this automatic and sustainable system. First, we will use generative artificial intelligence techniques to develop sustainable cementitious composites that include waste-derived components. Second, using computational simulations and laboratory-scale prototypes, we will optimize the developed cement composites for use in 3DCP. Third, we will design a cable-driven robot system for 3DCP with newly proposed control algorithms for automatic, reliable, and high-quality printing processes. Fourth, we will develop an automated 3DCP control system with real-time monitoring and feedback, as well as digital twins of the 3DCP system that can interact with computational models. Finally, we will assess the social, economic, and environmental factors that may affect the sustainable implementation of the developed system by the construction industry in Hong Kong. The overall objective of this proposed project is to furnish the construction sector in Hong Kong with a complete system of automated, robust, and cost-effective 3DCP technology that uses sustainable building materials and produces minimal waste. This project will contribute to alleviating housing and labor shortage, saving landfill space, and reducing carbon emissions, in Hong Kong, the Greater Bay Area (GBA), and internationally.