2850: Multi-Pathogen Respiratory Virus Modelling for Health Demands
This project will investigate current methodology for forecast models of respiratory disease for healthcare demand, to explore new and emerging modelling methods, and to progress methodology to accurately and robustly forecast 2 to 6 weeks of healthcare demand burden.
About the project
What the project aims to do
The overall aim of this project is to investigate current methodology for forecast models of respiratory disease for healthcare demand, to explore new and emerging modelling methods, and to progress methodology to accurately and robustly forecast 2 to 6 weeks of healthcare demand burden specifically from respiratory viruses across the UK.
Why this project is important
Respiratory viruses, like the common cold or flu, are still a big problem for healthcare systems around the world. Infectious disease modelling (using computers to understand how diseases spread among people, predict future outcomes, and make decisions for controlling and preventing their spread) can be used to help create these predictions.
This project aims to find better ways to predict how many people will need healthcare because of respiratory viruses. It will use different models to make more accurate forecasts and help hospitals plan for times when lots of people might get sick. It’s also important to figure out which respiratory viruses are causing the most problems, so we can focus on treating them. This project will look at how many people in the local area get tested for these viruses and compare it to how many people end up in the hospital. This will help Imperial College predict how many people might need hospital care in the future based on the number of people getting tested. It could also give Imperial College clues about how COVID-19 will affect other respiratory viruses after the pandemic.
Who the data is about
The data is about either:
- patients diagnosed with one or more seasonal respiratory viruses or COVID-19
- patients tested for influenza
It also includes:
- all ages
- gender (male or female)
- residency in England
It also includes:
- table 1: Admitted to a hospital from 1 January 2000 to 31 December 2022
- table 2: Discharged from hospital between 1 January 2000 to 31 December 2022
- table 3: Tested for influenza between 1 January 2000 to 31 December 2022
How the data will be used
Imperial College London will use historical data on hospital admissions to make predictions about the number of people who will be admitted due to respiratory viruses. To do this, the researchers will analyse the data over time and create a model that takes into account the seasonal patterns in hospital admissions and will also see if using specific data on the diseases themselves can help improve the accuracy of these predictions.
The researchers will break down the hospital admissions based on the specific respiratory viruses that caused them, such as influenza, SARS-CoV-2, and RSV. They will use data from surveillance systems that track these viruses to understand how they spread over time. By combining the number of cases of each infection, Imperial College will be able to predict the number of hospital admissions.
They will focus on developing a model that works similar to existing ones used in epidemiology. This model will consider factors like how infections spread and how the population’s immunity changes over time.
Additionally, the researchers will create simpler models for each infection that track the proportion of the population that is susceptible to the virus and those who have immunity, taking into account factors like infection rates and vaccinations. This will help us understand how infections and vaccinations affect immunity.
To test the accuracy of the models, Imperial College will train them using data from before the COVID-19 pandemic and then validate them using data from after the pandemic (2022 to 2024). They will analyse the data at different geographical levels, such as local authorities, NHS regions, and England as a whole. Imperial College will also consider whether certain model parameters should be the same for all regions or vary depending on the region. Additionally, I will explore models that take into account the correlation between neighbouring areas.
To improve the accuracy of the models, Imperial College will see if including external factors like vaccine coverage and effectiveness, Google Flu trends, weather data, and the timing of school holidays can help predict hospital admissions more accurately.
How often data is needed
One off release
How this project will benefit public health and the public
This type of analysis has not yet been conducted within the United Kingdom, and following the COVID-19 Pandemic, it will be important to understand how all the respiratory viruses effect the hospital system individually. Moving forward, if this model is successful, it may be used to better prepare hospitals and their staff for logistics, sudden rises in respiratory viruses, and future epidemics. The results of this project will be written up in the form of a PhD thesis, it will be presented at academic conferences, and submitted for publication. This work may contribute to improving public health surveillance capabilities for public health authorities, and lead to further policy making to combat National Health Service winter pressure.
Planned project outputs and communication
The research will appear in:
- peer reviewed scientific journals
- conference presentation
Lawful processing of personal and special category personal data
The data needed for this project is not personal data.
Legal basis for using personal data (Article 6)
Not applicable
Legal basis for using special category personal data (Article 9)
No applicable
Common law duty of confidentiality
The data needed for this project is not confidential patient information.
How is the duty of confidentiality set aside
Not applicable
National data opt-out
The opt-out preferences will not be applied.
This is because the national data opt-out does not apply to anonymised data.
Digital Object Identifier
Not assigned to this release.