Research

The Laboratory of Multiscale Host–Pathogen Dynamics, led by Dr. Kaiyuan Sun, applies mathematical and computational modeling to uncover key mechanisms in respiratory virus transmission and immune protection. The lab integrates approaches from epidemiology, immunology, bioinformatics, and complex systems theory to bridge molecular, individual, and population scales. Current research focuses on respiratory viruses such as influenza, RSV, and SARS-CoV-2, with the goal of supporting vaccine design, epidemic control, and pandemic preparedness through mechanistic insight and predictive modeling.

Direction 1. Understanding immune correlates of protection

The lab integrates serological, clinical, and epidemiological data to identify immune markers associated with protection against infection and transmission. By modeling immune memory, imprinting, and antigenic evolution, the team aims to establish a quantitative framework to support the evaluation and development of next-generation vaccines.

Direction 2. Linking within-host dynamics, population-level transmission, and viral evolution

Cross-scale models are developed to connect viral replication and immune responses at the individual level to transmission and evolutionary dynamics at the population level. These models help elucidate key processes such as antigenic drift, immune imprinting, and pathogen adaptation.

Direction 3. Integrating mechanistic modeling with machine learning to optimize epidemic control strategies

The lab builds data-driven platforms that integrate epidemiological and behavioral data to assess the multiscale impacts of vaccination, non-pharmaceutical interventions, and reactive behaviors. By incorporating game theory and multi-agent reinforcement learning, the team explores adaptive and intelligent epidemic response strategies under real-world constraints.

Direction 4. Applying protein language models to forecast viral evolution and improve strain prediction

Deep learning–based protein language models and molecular evolutionary analyses are used to predict adaptive mutations in viral genomes. This approach bridges molecular-level sequence features with epidemiological modeling to enhance the forecasting of variant emergence and transmission potential.

Direction 5. Developing integrated platforms for early warning and risk forecasting of infectious disease threats

The lab combines real-time surveillance data, pathogen genomics, and human behavioral information to build dynamic modeling platforms for risk assessment and trend forecasting. These systems aim to support early detection of emerging pathogens, monitor epidemic dynamics, and strengthen global health preparedness.