A new MOnitoring Outbreaks for Disease surveillance in a data science context (MOOD) Horizon 2020 study published this week offers an insightful look at how disease modelling and data use changed throughout the COVID-19 pandemic. The research, led by Dr Esther van Kleef, Technical Consultant for the World Health Organization (WHO), with WorldPop’s Professor Andy Tatem and Dr Shengjie Lai and others was conducted across Europe and examined how scientists’ modelling priorities evolved and how effectively they worked with decision-makers from 2020 to 2022. The findings shed light on what worked, what didn’t, and what needs to improve before the next global health crisis.
At the start of the pandemic, modelling teams raced to understand the virus itself, how fast it spread, how deadly it was, and how many people might be infected. As the crisis unfolded, focus shifted.
In the early phase (Jan–Jun 2020), most efforts went into understanding COVID-19’s basic transmission dynamics. By the mid-phase (Jul 2020–Jun 2021), attention turned to evaluating the impact of non-pharmaceutical interventions (NPIs) like lockdowns, mask mandates, and travel restrictions. In the later phase (from Jul 2021), as vaccines became available, models were increasingly used to assess how vaccination campaigns affected infection rates and immunity.
Throughout all three phases, one thing remained constant: the need to understand how the virus spread through different communities and regions. While COVID-19 generated an unprecedented amount of data, the study revealed that data gaps were a recurring problem. Modellers relied heavily on traditional surveillance data (case numbers, hospitalizations, and death counts) supplied by public health authorities. But for more complex questions, such as how social behaviour or population movement affected transmission, key information was often missing.
More than half the 66 studies analysed lacked critical data sources such as:
- Social contact data (how people interacted in daily life)
- Behavioural surveys (attitudes toward restrictions or vaccines)
- Serological data (levels of immunity in the population)
The main reasons? These data were often never collected, took too long to access, or were restricted by privacy or ethical rules. Even when available, data were not always standardised or detailed enough to reveal differences by age, sex, or health conditions.
Interestingly, about a quarter of the studies benefited from data provided by private companies, showing how partnerships beyond government sources played a key role. The study also highlights how scientists and decision-makers learned to work together under pressure.
In most cases (58 out of 66 studies), the modelling directly informed public health decisions or improved officials’ understanding of the situation. Early in the pandemic, collaboration was often direct and informal, but over time it became more structured, involving advisory committees and official working groups.
Research focused on forecasting outbreaks or assessing interventions had the greatest impact on policy, with many cited in government documents and reports. However, the rush to provide rapid results came with trade-offs. Fewer than half of the studies made their code publicly available, as researchers prioritised speed over open-source documentation during the emergency.
The study’s authors argue that to prepare for the next pandemic, we need to build sustainable, standardised, and transparent data systems, ones that go beyond case counts to include richer demographic and geographic information.
They call for:
- Better data-sharing protocols for both traditional and new data sources.
- Stronger collaboration networks between scientists, governments, and private organisations.
- A global culture of openness in sharing data and modelling tools.
As Dr van Kleef and her colleagues emphasise, future preparedness depends not only on new technologies but on how well we share information, trust one another, and act on the evidence.
The MOOD project was a Horizon2020 European Union (EU)-funded project (Jan 2020 to Dec 2024), that brought together partners from academic, research, public health and animal health institutions.
Image credit: Man wearing mask by Jeyaratnam Caniceus via Pixabay
We’re trialling the ‘Deep Dive’ audio summary feature of Google’s NotebookLM. This feature uses AI to create a podcast-like audio conversation between two AI-derived hosts that summarise key points of documents - in this case the Modelling practices, data provisioning, sharing and dissemination needs for pandemic decision-making paper listed below.
As Google acknowledge that NotebookLM outputs may contain errors, we have been careful to check, edit and validate this audio.
Please contact us to let us know what you think.
Music: My Guitar, Lowtone Music, Free Music Archive (CC BY-NC-ND)
