Theme 1: Understanding, Preventing and Treating Diseases
Project Title: Microbiota in Gastrointestinal Cancers: Mechanistic Investigation and Clinical Applications
Project Coordinator: Prof Jun YU (CUHK)

Abstract

Gastrointestinal (GI) cancers, including gastric cancer (GC) and colorectal cancer (CRC), are among the killers in Hong Kong and the Chinese Mainland. The clinical outcomes of GI cancers remain unsatisfactory due to their late diagnosis, unresponsiveness to cancer therapy, and the emergence of therapeutic resistance. Research in this area is essential to uncover clinically translatable solutions to save many lives.

Emerging research suggested that the gut microbiota – the collection of microbial organisms (including bacteria, viruses, fungi, etc.) living in our gut – critically contributes to GC and CRC development, and holds significant potential for clinical application. The project team is at the forefront of research on GI cancer microbiota. The team was the first to document the changes in gut microbiota during GC and CRC progression, and identified new “bad” bacteria that could promote cancer development, as well as “good” bacteria that guard against GI cancer. Moreover, the team pioneered the application of stool bacteria as biomarkers (measurable indicators) for non-invasive GI cancer detection, and probiotics (“good” bacteria) for cancer prevention.

Building on the international-leading expertise of the team, the mission of this project is to (1) deepen our understanding of gut microbiota in GI cancer pathogenesis and (2) translate the new knowledge into microbiota-based prognostic (predicting disease progression, risk, and overall survival in patients) and therapeutic tools for GI cancers.

The project encompasses four objectives involving basic discovery and translational research. In the discovery phase, we will first utilise state-of-the-art technological platforms to map changes in the gut microbiota in GC and CRC at a single-cell resolution in order to identify the key microbes that promote or suppress GI cancer formation. Second, we will extensively determine the function of the identified microbes and elucidate their molecular mechanisms of action in relevant models of GI cancers. These discoveries will provide the foundation for the downstream translational research, including formulation of microbial biomarker panels with artificial intelligence (AI)-assisted algorithms for accurately predicting whether GI cancer patients will respond to cancer therapy; and development of probiotics, postbiotics (bioactive compounds made by “good” bacteria), and phages (viruses that infect and replicate within bacteria but not human cells) as novel therapies for GI cancer prevention, as well as boosting the effectiveness of current cancer therapies.

In summary, this is a multidisciplinary project bridging basic and translational research in GI cancers which will generate valuable intellectual property with translational and economic values, nurture scientific talents, and establish Hong Kong as an international hub for the microbiome-based biotech industry.

Theme 1: Understanding, Preventing and Treating Diseases
Project Title: Next-Generation Orthopaedic Implants: R&D and Clinical Translation with Multi-functional Properties of Fixation, Regeneration, and AI-based Therapy Monitoring
Project Coordinator: Prof Ling QIN (CUHK)

Abstract

Population ageing is a global challenge. The number of individuals over 65 years old is predicted to exceed 1.6 billion worldwide, with Hong Kong as the world’s “oldest city”, where 40.6% of the population is projected to be over aged 65 by 2050. Ageing is associated with osteoporosis, fragility fractures, and osteonecrosis, leading to joint deformity, disability, costly joint replacement procedures and complications associated with high mortality. We have established a professional industrial platform for the manufacture of magnesium (Mg)-containing orthopaedic implants, third-party testing, and pre-registration. We focus on the clinical translation of novel orthopaedic implants for the treatment of osteonecrosis of the femoral head (ONFH) and femoral neck fracture (FNF), targeting not only Hong Kong but also national and global markets.

To achieve these objectives, this TRS project will focus on three aspects:

1) Clinical Translation of Mg-containing innovative orthopaedic implants for ONFH and FNF by obtaining the required third-party test reports and collaborating with a contract clinical research organization (CRO) to conduct multi-centre RCTs, and to obtain the Class III Medical Device Certificates from the Chinese National Medical Products Administration (NMPA). Over the next 5-6 years, our efforts will realise next-generation clinical orthopaedic implants to enable faster, better healing and avoid revision surgery, transitioning from current anatomical reduction to functional recovery. We will also prepare for the U.S. Food and Drug Administration (FDA) “Breakthrough Device” certification and the 510(k) premarket application. This initiative aims to generate two Class III Medical Device Certificates from NMPA, followed by License Listing in Hong Kong and 510(k) clearances with the FDA, and exploration for CE of Europe;

2) Mechanism Insights: We will investigate the cellular and molecular mechanisms involved in tissue regeneration following Mg implantation, using cutting-edge technologies such as tissue clearing, non-diffracting light sheet imaging, and single-cell spatio-temporal transcriptomics to understand tissue crosstalk in relation to Mg degradation and fracture healing enhancement. This basic research part will generate top-tier academic publications, strengthen the theoretical foundation, and support the “Breakthrough Device” application with both NMPA and FDA; and

3) R&D of Smart Orthopaedic Implants that incorporate artificial intelligence (AI) to monitor fracture healing progression by fitting the healing curve and patient data. These smart implants will monitor the healing stage and quality and warn of potential delayed healing or non-union for timely clinical intervention to advance orthopaedic care and improve therapeutic effect and patient outcomes through our multidisciplinary innovations.

Theme 2: Developing a Sustainable Environment
Project Title: Preparation of Novel Unconventional Cu-based Catalysts for CO₂ Conversion towards Scalable Production of Value-added Chemicals
Project Coordinator: Prof Hua ZHANG (CityU)

Abstract

Carbon dioxide (CO₂) emitted from burning fossil fuels is one of the main causes of climate change. Now renewable energy sources, such as solar and wind power, are becoming increasingly common. However, their intermittent nature limits the continuous output of electricity. One promising solution is to convert CO₂ into valuable chemicals and fuels by using electricity from renewable sources. This strategy not only helps reduce carbon emissions but also stores renewable energy as useful products. Copper has a unique ability as a catalyst to convert CO₂ into more complex and high-value multi-carbon products, such as ethylene, ethanol, acetate and propanol. These products have much higher economic values than the basic chemicals like carbon monoxide or methane. However, current copper-based catalysts still face major challenges, such as low efficiency, poor stability under real operating conditions, and low activity for selectively producing the desired products when operating at high production rates required for industrial application.

This research project aims to overcome the aforementioned challenges by developing a new class of advanced copper-based catalysts with unconventional structures. We will design and create catalysts in the form of single atoms, dual atoms, special alloys, and composite materials supported on other surfaces. These new structures are expected to significantly improve the efficiency and selectivity when converting CO₂ into valuable chemicals. To accelerate the discovery process, we will combine machine learning, computer simulations and experimental testing to create a smart feedback system in which the computer helps predict and design promising catalysts, which are then synthesized and tested in the lab. The obtained results are fed back to the computer to improve future predictions and designs of new catalysts. In addition, we will use advanced real-time probing techniques to monitor how catalysts work during CO₂ conversion, thereby helping us to understand the relationship between the catalysts’ structures and their performance. The key of this project is to improve the overall energy efficiency of the system. Normally, a large amount of energy is wasted at the other electrode for a reaction that produces oxygen. We plan to replace this with more useful reactions, such as converting alcohols, acids, or biomass into valuable chemicals. This “paired electrolysis” approach can make the entire process more economical. Ultimately, the project targets the development of catalysts and prototype electrolyzers that can operate at high production rates with good stability over long periods. Our goal is to create a practical technological pathway that supports Hong Kong’s target of achieving carbon neutrality by 2050, while also contributing to global efforts in sustainable energy and carbon management.

Theme 2: Developing a Sustainable Environment
Project Title: Intelligent Artificial Photosynthesis
Project Coordinator: Prof Lianzhou WANG (PolyU)

Abstract

What if carbon dioxide were no longer only an emission burden, but could be wisely used as a resource for green fuels using sunlight? Our mother Nature already gives us a powerful example. Green plants use sunlight every day to convert carbon dioxide and water into energy-rich substances that sustain life for millions of years.

This program, “Intelligent Artificial Photosynthesis”, aims to learn from nature and build a new research platform for recycling carbon dioxide using sunlight. It will combine artificial intelligence, advanced materials, automated experiments and solar fuel reactor design to develop a closed-loop “Design–Make–Test–Learn” system. Artificial intelligence will accelerate the design and selection of new catalytic materials; automated systems will enable rapid preparation and testing of material; in-situ and operando facility will help better understand how the reaction works; and the results will be fed back to guide the next round of improvement.

The main goal is to use sunlight to convert carbon dioxide and water into liquid fuels such as green methanol. Methanol is a liquid, which is easier to store, transport and use than hydrogen gas. The sustainable production of methanol will support future low-carbon shipping, sustainable fuel supply and green chemical production. The team will also develop modular photothermal panel reactors and work toward high efficiency, high selectivity and long-term stable operation, exploring a pathway from laboratory research to future application.

Theme 3: Enhancing Hong Kong’s Strategic Position as a Regional and International Business Centre
Project Title: Managing AI Risk for Businesses in Hong Kong
Project Coordinator: Prof Feng TIAN (PolyU)

Abstract

Artificial intelligence (AI) is rapidly changing how businesses in Hong Kong operate. It brings efficiency gains and new opportunities, but also new risks. For example, employees may use generative AI tools without approval, AI-generated content can spread false information, and some jobs may be reshaped or even displaced. This project will develop a practical, evidence-based AI risk management framework to help businesses in Hong Kong adopt AI safely and with confidence.

The research consists of four parts. First, we will build a clear picture of how AI is used by companies and workers across industries in Hong Kong. We will also develop new methods to detect the unauthorised use of generative AI (GenAI) in corporate reports and analysts’ reports, and examine how such use affects the quality of information available to investors.

Second, we will assess the risks that come with corporate AI adoption, focusing on compliance and operational risks. We will also study how GenAI facilitates the spread of misinformation and how governments can regulate such risks, for example through rules that require AI-generated content to be clearly labelled, in order to reduce AI-related misinformation and strengthen investor confidence in Hong Kong’s stock market.

Third, we will study how AI reshapes the job market in Hong Kong, including changes in demand across occupations, effects on employment at firms, and the movement of talent across borders.

Finally, we will conduct annual forward-looking studies with our industry partners to forecast emerging AI risks. Drawing all these findings together, we will deliver an AI risk management framework tailored to Hong Kong, together with an open-access AI risk management service for local firms.

Theme 4: Advancing Emerging Research and Innovations Important to Hong Kong
Project Title: SynapseCity: AI-Driven Urban Emergency Response and Rescue
Project Coordinator: Prof Jianping WANG (CityU)

Abstract

As a highly dense city, Hong Kong’s roads, drainage, electricity, telecommunications, and transport systems are tightly interconnected. When typhoons, rainstorms, landslides, or fires occur, a localized failure can easily trigger cascading effects: road flooding may delay fire trucks and ambulances, while a tunnel fire may also damage nearby power cables or communication lines. In large-scale incidents, agencies such as government departments, public utilities, and transport operators must spend substantial time manually verifying data, confirming equipment, and coordinating routes, which can delay action as conditions change rapidly. Meanwhile, although different parties hold useful field data and predictive models, these raw data cannot be centralized into a single database because of privacy, commercial confidentiality, and operational security concerns.

To fill this communication and coordination gap, this project proposes SynapseCity, an AI-driven open platform whose key feature is that agencies do not need to hand over their raw data. Each authorized organization is equipped with a dedicated “AI Agent” that only accesses its internal data and understands its mandates and tools. Under privacy protection, these AI agents exchange judgments and early warnings in the back end, rather than sharing raw data directly. This allows front-line commanders to grasp the overall situation earlier, identify where risks are rising, what cross-departmental resources are available, and what needs further verification.

SynapseCity puts security and privacy first. Sensitive internal data, operational logic, and specialized models, such as drainage hydraulic simulations and logistical routing tools, remain on the original organizations’ servers. Data owners can define rules governing what their AI agents may read and share. Since communication networks may become congested or partially disrupted during disasters, the project develops resilient emergency communication support technologies to ensure that AI agents can still exchange critical messages under extreme conditions. This enables public and private partners, such as bus companies and power utilities, to contribute domain expertise to rescue collaboration while minimizing operational and legal risks.

SynapseCity’s core breakthrough is resolving a long-standing tension in urban crisis management: agencies must protect data privacy, proprietary knowledge, and operational security, while major disasters require immediate collaboration and shared judgment. By using decentralized AI agents, SynapseCity turns cross-sector intelligence from government departments, utilities, transport operators, and private partners into actionable, explainable, life-saving response and recovery plans, reducing reliance on repeated communication and multi-party confirmation across fragmented data and processes. This gives front-line teams operational agility that fragmented planning models could not achieve. Although designed for Hong Kong’s dense hillsides, coastal exposure, busy transport corridors, and highly interdependent infrastructure, its modular, multi-agent, privacy-first architecture can scale to other high-density cities. In the long term, SynapseCity offers a global model for moving from distributed operations to integrated collaboration, and from reactive response to resilient governance, using decentralized AI to safeguard critical infrastructure and citizens’ safety.

Theme 4: Advancing Emerging Research and Innovations Important to Hong Kong
Project Title: 3D-Integrated Power Delivery Technology for Advanced AI Chip Modules
Project Coordinator: Prof Han WANG (HKU)

Abstract

As the worldwide demand for AI data center infrastructure grows rapidly, their power consumption is increasing exponentially, creating an urgent need for more efficient and compact power delivery technologies. Traditional power architectures for AI datacenters face fundamental limitations in efficiency, thermal management, and integration density. This project proposes a holistic 3D-integrated power delivery solution that combines gallium nitride (GaN) power semiconductor devices, 3D integration, miniaturized passive components, advanced power electronics, and novel cooling technologies. Key innovations include monolithic and heterogeneous integration of GaN/Si CMOS and GaN power stages in 3D-interposer architectures, a five-fold reduction in passive component size through novel materials and 3D structuring, and co-design of power and thermal management using SiC interposers with microfluidic cooling.

The project aims to enable multi-kW/mm² power delivery density for AI datacenter chip modules, reduce power conversion losses by more than 50%, and achieve a paradigm shift in energy efficiency for AI and high-performance computing. Critical for the future AI datacenter infrastructures both locally and worldwide, the proposed research is expected to deliver key technological know-how with strong economic value and also establish close collaboration with leading global semiconductor manufacturers. The research outcome can chart a pathway to enable sustained gains in computing performance alongside transformative improvements in energy efficiency and sustainability of future AI datacenters.

Theme 4: Advancing Emerging Research and Innovations Important to Hong Kong
Project Title: Development and Assessment of Human Capacities for Productive Human–AI Interaction
Project Coordinator: Prof Dragan GASEVIC (HKU)

Abstract

The growing ubiquity of generative artificial intelligence (AI) tools such as ChatGPT is reshaping how young people learn, solve problems, and create artifacts. Schools and families are adopting these tools without evidence about their effects on learning. While AI can improve productivity and performance, it also raises concerns about overreliance and work that appears polished but masks weak understanding. Such outputs can create an illusion of competence, potentially leading learners to overestimate what they know and reducing the likelihood that they will question AI responses or monitor their understanding. Recognising and responding to these misleading signals depends on learners’ regulatory capacity, which enables them to monitor understanding, judge quality, and decide when to rely on AI and when to sustain their own effort. The need is especially urgent for learners aged 9–15 because this period is particularly important for the development of regulatory capacity and coincides with growing independent use of digital technologies including AI.

The mission of this exploratory TRS project is to establish the foundations needed to understand and strengthen the regulatory capacity that supports productive human–AI interaction. The exploratory work will be conducted in partnership with Hong Kong schools and will involve learners aged 9–15, teachers, and parents. We will a) conduct surveys with students, teachers, and parents, together with systematic literature and policy reviews, to understand current practices, identify practical needs, and inform future research and practice; b) use a digital learning environment to collect preliminary evidence about how learners regulate their interaction with AI while working on problem-solving and co-creation tasks; c) establish the feasibility of using digital data recorded by the learning environment to measure regulatory capacity during learner–AI interaction; and d) conduct co-design sessions with teachers and students to develop learning tasks, teaching approaches, assessment practices, and requirements for technology support.

The exploratory project will draw on an interdisciplinary team with expertise in education, psychology, assessment, learning analytics, artificial intelligence, human–computer interaction, and ethics. The team’s experience in school-based research, digital learning environments, educational assessment, and co-design will enable the project to identify patterns of regulatory capacity, test their assessment through authentic learning tasks, and prepare the measures, tasks, technology, and school partnerships required for a future longitudinal programme. The project will inform educational practice, assessment, technology development, and policy while strengthening Hong Kong’s capacity to lead research on human learning and human–AI interaction.