Health - Personalised Healthcare

Epidemic preparedness: the power of data science

Date:
Changed on 20/12/2024
On the occasion of International Epidemic Preparedness Day, the Sistm project-team from the Inria Center at the University of Bordeaux is demonstrating the crucial importance of data science in epidemiology. By transforming the vast quantities of data from vaccine trials into usable information, this team is helping to speed up vaccine development, optimize clinical trials and devise strategies for controlling epidemics.
© Inria / Photo C. Morel

The crucial role of data science in epidemiology

At the heart of the research carried out by the Sistm project team, a joint venture between Inria, Inserm and the University of Bordeaux, is the provision of expertise in data science, enabling the vast quantities of data from vaccine trials to be transformed into usable information. “At Sistm, we analyze data from clinical trials conducted by our partners. This enables us not only to validate or better understand the mechanisms of action of vaccines, but above all to quantify their efficacy and, subsequently, to propose optimal strategies for new trials”, explains Mélanie Prague, head of the Sistm project team.

This team works on all aspects of infectious diseases, from accelerating vaccine development to clinical trial design and recommendations on epidemic control strategies. “Historically, the team worked a lot on HIV. Every year, the WHO (World Health Organization) publishes a list of potentially pandemic pathogens. We are working more and more on these emerging diseases, such as Ebola, SARS-CoV-2, or the Nipah virus. But we also have other applications,” explains Mélanie Prague.

To achieve this, scientists work closely with the Vaccine Research Institute, where biologists provide data on the pre-clinical animal and clinical phases of vaccines. The team uses mathematical models based on differential equations to simulate the biological mechanisms of vaccines, and develops advanced statistical methods to identify significant signals in the midst of complex, multi-sourced, high-dimensional data.

Helping to accelerate vaccine development

Traditionally, it takes between 10 and 15 years to develop a vaccine, and only 16% of candidates that reach the clinical phase manage to obtain marketing authorization. With the growing urgency of pandemics and the rapid emergence of new infectious diseases, it's essential to accelerate vaccine development in order to protect the world's populations more effectively and more rapidly.

The Sistm project team's innovative approach to numerical simulation makes it possible to virtually test immune reactions, selecting the most promising vaccine responses before moving on to clinical trials. This method, mainly used upstream of marketing authorization, is crucial for understanding how vaccines work and determining which versions are the most effective. By using statistical models to predict human reactions from preclinical data, this team plays a key role in the “bridging” process. This concept, recognized by health authorities, is essential in the development of drugs and vaccines. Bridging makes it possible to transfer efficacy and safety data obtained in one population group (for example, animal models such as monkeys, or a human study sub-population) to another group, without having to repeat all the clinical trials for each new population. This approach, while respecting rigorous evaluation criteria, speeds up the process of approving treatments for different patient groups, while reducing the costs and time needed to bring them to market. As Mélanie Prague points out, “this makes it possible to optimize resources and adapt choices more quickly by adopting the most promising strategies.”
 

 

© Inria / Photo B. Fourrier

Statistics to design clinical trials

One of the key contributions of the Sistm project-team lies in the design of optimized clinical trials. In collaboration with the Vaccine Research Institute, the team develops predictive statistical methods to generate new hypotheses from the data, thereby maximizing trial efficiency. When a clinical trial is launched, the first step is to manage the data received and carry out statistical analyses. These trials produce an abundance of complex data, often in large dimensions. “Numerous markers are measured to track biological dynamics. Between gene transcription data, different cell types and antibody-related information, there can be over 20,000 variables measured at five or six visits for just thirty or so patients. The analysis of this longitudinal data repeated over several individuals represents a challenge due to the high dimensionality of the data and the small number of subjects,” explains Mélanie Prague.

To simplify this complexity, the Sistm project team follows a multi-stage approach. First, in the “data integration” stage, scientists seek to pinpoint important information using different methods. For example, they may identify genes that are expressed differently between groups, or develop techniques to automatically count cells, a task once done “by hand via software” by biologists, making the process faster and more reliable. The next step is to model the mechanisms involved. “In this stage, we'll work with inverse problems: based on the data collected, we'll try to define the most likely immune model that may have generated the data. These models can then be used to make predictions based on the available data”, continues the researcher.

This approach generates new hypotheses, which can then be used to design new clinical trials. In this way, the Sistm project-team's approach integrates data management, statistical analysis, mechanism modeling and hypothesis generation, forming a continuous cycle of optimization and innovation in clinical research.

Optimizing infectious disease control

The Sistm project team's expertise also extends to epidemic control. During the Covid-19 pandemic, for example, the team focused on modeling the spread of the epidemic and evaluating non-pharmaceutical control strategies, such as containment. The aim was to identify and propose optimal strategies for containing the epidemic, but also, after the fact, to enable better preparation for future epidemics, while taking into account the economic, psychological and societal costs engendered by these strategies. “The aim is not to propose a hypothetical method that “might work”, but to evaluate and recommend optimal strategies, taking all options into account. The aim is to present predictions adapted to each situation, based on data from experiments already carried out, on which healthcare decision-makers can base their decisions,” emphasizes Mélanie Prague.

In the case of Covid-19, scientists from the Sistm project team, in collaboration with McGill University in Canada, used a mathematical model based on French public data to estimate the effects of containment, curfew and vaccination measures. Their results, published in early 2024 in the journal Epidemics, show that containment and curfews significantly reduced transmission of the virus, with an 84% reduction for the first containment. Without vaccination, there would have been 159,000 additional deaths and 1.48 million hospitalizations in France. These simulations underline the importance of extreme reactivity in vaccine deployment and decision-making to limit the impact of epidemics, providing guidance to decision-makers should a new pandemic occur.

 

© Inria / Photo B. Fourrier

The Sistm project-team at the Inria Centre at the University of Bordeaux illustrates how the integration of data sciences can transform the fight against epidemics. From accelerating vaccine development to designing optimized clinical trials and controlling infectious diseases, the Sistm project-team continues to push back the frontiers of innovation in digital public health. Through its collaborations and methodological advances, the team is positioning itself as a key player in preparing for and responding to future epidemics.