Hong Kong Introduces AI Model for Predicting Passenger Travel Demand

In a significant advancement for urban transportation management, Hong Kong's principal railway operator, MTR Corporation Limited, has launched a cutting-edge ridership prediction model aimed at optimizing passenger distribution and travel demand amidst disruptions. This initiative was discussed by May Tso, the Senior Technology Development Manager at MTR, during a panel at the UITP Summit in Hamburg on June 22, 2025. Tso emphasized that this model represents an unprecedented approach to operations planning, particularly beneficial during unexpected events that could disrupt normal services.

The ridership prediction model, which simulates passenger flows throughout the city's public transport network, integrates data from government surveys and employs machine learning techniques trained on over one billion transactions. This robust model allows for accurate predictions within a margin of 10%, enabling operators to make informed decisions regarding passenger routing during peak and off-peak times, as well as accommodating the travel behaviors of diverse demographic groups, including students and the elderly.

"The model’s capability includes not only metro operations but extends to trams, buses, minibusses, and ferries operating in Victoria Harbour," Tso noted. The application of this technology serves as a digital twin of the city’s public transportation infrastructure, providing a comprehensive overview of passenger distribution at any given time.

MTR Corporation's collaboration with a local university has been pivotal in transforming theoretical research into practical applications within the railway sector. Through this partnership, they have effectively harnessed years of academic research to devise a model that enhances operational efficiency.

Prior to the implementation of this model, operators at the central traffic control center faced significant challenges during service disruptions, often having to shift focus from regular operations to incident management. As Tso explained, the new model aids in resource planning and crowd management, diminishing the manual effort required during crises and allowing for a quicker recovery process. This innovation not only benefits MTR but also enhances the overall travel experience for commuters in Hong Kong.

Looking ahead, the implications of this technology extend beyond immediate operational improvements. The ability to predict passenger demand accurately could lead to more sustainable transport strategies, reducing congestion and improving service reliability. Furthermore, as urban areas around the world grapple with similar challenges, Hong Kong's proactive approach may serve as a model for other cities seeking to integrate AI solutions into their public transport systems.

The successful deployment of the ridership prediction model reflects a growing trend in urban mobility where cities leverage technology to enhance resilience and responsiveness in public transportation. As demonstrated in other global cities facing similar challenges, the strategic use of data and AI could become essential in reshaping the future of urban mobility and ensuring the seamless movement of people in increasingly congested environments.