Supervisory team
Nagasaki University
- Lead supervisor: Fuminari Miura ([email protected])
- Third supervisor: Lina Madaniyazi ([email protected])
- Additional supervisor: Masahiro Hashizume ([email protected])
LSHTM
- Second supervisor: Alexis Robert ([email protected])
The project
Infectious disease surveillance provides information on the current or past transmission dynamics of pathogens within populations. However, transmission is driven by factors that are often not included in surveillance datasets, such as on people’s behaviour (e.g. contact patterns, immunity in the population, mobility), pathogen characteristics (e.g. mode of transmission, infectiousness, generation time), environmental determinants (e.g. climate, urban-rural settings), and stochasticity. Epidemic modelling can integrate multiple data streams to reconstruct past transmission and anticipate future outbreak risk. A central challenge is to identify which external signals contain reliable information about changing transmission conditions and can improve prospective epidemic forecasting.
This PhD will develop modelling frameworks for forecasting local transmission risk by integrating environmental and other 'alternative' data directly into time-varying transmission models. The first component of the project will be centered on methodological development. The student will extend a semi-mechanistic model that decomposes reproduction numbers into susceptible depletion and external drivers (te Beest et al., 2013), jointly modelling latent infections, endogenous within-epidemic dynamics and nonlinear distributed-lag effects of climate. The framework will then be generalised beyond climate to mobility, behavioural and economic signals, assessing whether multiple noisy data streams can act as proxies for latent transmission processes. The performance of these methodological development will be assessed using weekly European influenza-like illness (ILI) data available through ERVISS/RespiCast, allowing rigorous 1-4-week probabilistic forecast evaluation against established baselines and models.
The second component of the project will apply and adapt this framework to pathogen-specific data streams such as influenza and respiratory syncytial virus (RSV) in Japan, using prefecture-level surveillance data. By integrating local demographic characteristics (such as age distribution, socioeconomic inequalities, urban / rural status) into the model previously developed, we will aim to identify which prefectures are at highest outbreak risk. These analyses will examine which mechanisms explain changes in epidemic onset and seasonality, including climate, susceptible depletion or replenishment, childhood mixing, subtype circulation and international seeding. Spatial transmission modelling using regional connectivity will be an optional extension where data support it.
The work will combine hierarchical statistical modelling, distributed lag nonlinear models, renewal and other stochastic processes, simulation, latent-state methods and prospective forecast evaluation. The overall objective is to produce an interpretable, flexible, and reusable framework for translating heterogeneous external data into epidemic intelligence.
References
- Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med. 2010;29:2224-2234. doi:10.1002/sim.3940.
- te Beest DE, van Boven M, Hooiveld M, van den Dool C, Wallinga J. Driving Factors of Influenza Transmission in the Netherlands. Am J Epidemiol. 2013;178:1469-1477. doi:10.1093/aje/kwt132.
- Bhatt S, Ferguson N, Flaxman S, Gandy A, Mishra S, Scott JA. Semi-mechanistic Bayesian modelling of COVID-19 with renewal processes. JRSS A. 2023;186:601-615. doi:10.1093/jrsssa/qnad030.
- Park SW, Holmdahl I, Howerton E, et al. Interplay between climate and childhood mixing can explain a sudden shift in RSV seasonality in Japan. Nat Commun. 2025;16:11385. doi:10.1038/s41467-025-66184-y.
- Wagatsuma K, Feurer D, Yu W, et al. The joint impact of temperature, humidity, and air pollution on COVID-19 incidence: a multi-country time-series study in 439 cities. Environ Int. 2026;208:110090. doi:10.1016/j.envint.2026.110090.
The role of LSHTM and NU in this collaborative project
Nagasaki University
Dr Fuminari Miura will lead the project and have primary responsibility for the overall scientific direction, day-to-day supervision, integration of the methodological and applied components, and coordination across the supervisory team. He will lead the development of the semi-mechanistic epidemic modelling framework and its application to infectious-disease surveillance and alternative data.
Prof. Lina Madaniyazi (Nagasaki University) will provide expertise in environmental epidemiology, distributed-lag and environmental time-series modelling, and climate-health interpretation, and will guide the environmental-data component of the project. Prof. Masahiro Hashizume (Nagasaki University) will contribute as a senior advisor, particularly on environmental epidemiology, climate-sensitive infectious diseases, and the broader public-health interpretation and positioning of the work.
LSHTM
Dr Alexis Robert will provide complementary expertise in infectious-disease dynamics, semi-mechanistic and renewal modelling, probabilistic forecasting, and heterogeneous transmission risk. Dr Robert will co-supervise the methodological and forecasting components and help develop the connection between external drivers and transmission dynamics.
The student will be primarily based at Nagasaki University, consistent with the NU lead supervisor, with opportunities for research periods at LSHTM. Depending on the final project structure and practical arrangements, the student may also spend research periods in the Netherlands, where Dr Miura is primarily based, to work closely on model development, forecasting, and alternative-data analyses. The supervisory team will jointly define the research programme, review analyses, code and manuscripts, and hold regular cross-institutional meetings to ensure that the project is developed as a collaborative LSHTM–NU PhD.
Particular prior educational requirements for a student undertaking this project
Essential: a strong quantitative Master's-level background in statistics, biostatistics, applied mathematics, physics, computer or information science, quantitative epidemiology, data science, or a closely related field. Applicants should have foundations in probability, statistical inference, regression and mathematical modelling, together with strong programming skills in R, Python, or Julia.
Desirable: prior experience with infectious disease modelling or epidemiology, Bayesian or likelihood-based inference, dynamical systems, compartmental or renewal models, state-space models, environmental or spatiotemporal statistics, forecasting, or time-series analysis.
Skills we expect a student to develop/acquire whilst pursuing this project
- Infectious disease modelling, including renewal process, semi-mechanistic and compartmental models.
- Distributed lag non-linear models (DLNM), Generalised additive models (GAM) and environmental time-series analysis.
- Hierarchical/Bayesian or likelihood-based inference and latent-state modelling.
- Probabilistic short-term forecasting, calibration and forecast evaluation (e.g., WIS/log score).
- Integration of heterogeneous surveillance, climate, mobility, travel and economic datasets.
- Reproducible statistical programming in R, Stan or related tools as appropriate.
- Critical interpretation of predictive versus causal signals and communication of model results for public-health use.
- Optional: spatial epidemic modelling and connectivity/network-based extensions.