Using big data to prevent pandemics

Every year, around 2.6 billion people are affected by diseases originating in animals (zoonotic diseases). To prevent pandemics, it is essential to detect and characterize zoonotic diseases before an outbreak occurs, or at an early stage.

As part of the IAEA’s Zoonotic Disease Integrated Action (ZODIAC) initiative launched in 2020, the ZODIAC Respiratory Disease Phenotype Observatory will create a secure medical imaging repository to foster global cooperation on large scale data analysis of disease patterns, enabling the early detection of zoonotic diseases that could potentially cause pandemics.

The observatory will use artificial intelligence (AI), including machine learning and deep learning, to identify the patterns of respiratory diseases such as Middle East respiratory syndrome (MERS), severe acute respiratory syndrome (SARS), COVID-19 and pneumonia and detect the emergence of new variants.

“The IAEA’s ZODIAC Respiratory Disease Phenotype Observatory will play an important role in identifying the emergence of new infectious diseases around the world, monitoring their spread and facilitating the rapid development of AI models for treatment support,” said Professor Georg Langs, Head of the Computational Imaging Research Lab at the Medical University of Vienna, one of the project’s core laboratories. “Because it works with research institutions across the globe, the observatory will be able to analyse a much larger and more demographically diverse collection of data on respiratory diseases than previous studies.”

Medical imaging plays a crucial role in diagnosing and monitoring infectious diseases. However, images can be challenging to analyse because of their complexity.

The ZODIAC Respiratory Disease Phenotype Observatory will use radiomics, a method for extracting large scale imaging data, or big data, from medical imaging studies. Radiomics uses data characterization algorithms to identify disease findings, increasing diagnostic accuracy and aiding individualized therapy planning.

AI can complement radiomics by identifying disease patterns and anomalies in large volumes of data. These techniques can also be used to identify the patterns of emerging diseases, which can help to prevent outbreaks of novel diseases developing into pandemics.

In its first two years, the observatory will create a medical imaging repository and use it to develop and validate algorithms for analysing imaging data. It will evaluate studies conducted by 20 research institutions around the world, and will enable research groups to investigate and develop novel AI methods. The observatory’s AI-based surveillance for emerging diseases will function automatically, triggering an alert when algorithms detect a new pattern. It will enable the rapid comparison and evaluation of incoming data so as to identify the emergence of new diseases that could develop into pandemics and ensure a timely response.

The observatory will also analyse the demographics of new infectious respiratory diseases. By identifying disease characteristics and specific manifestations in medical imaging, the observatory can help to identify any clinical differences in the development of disease complications, based on factors such as age, sex, race, ethnicity, geographical region and pre-existing medical conditions.

Created through an IAEA coordinated research project, the ZODIAC Respiratory Disease Phenotype Observatory is supported by many partners who provide resources and tools in their respective areas of expertise.

Amazon Web Services (AWS), one of the project’s lead supporters, has awarded an AWS Grand Challenges grant for a cloud based server to support the observatory.

“We see this as an important investment in prevention to help protect human health globally,” said Chris Russ, Senior Solutions Architect at AWS. “By leveraging the cloud, the IAEA’s ZODIAC Observatory can spot emerging pandemics in real time and alert governments to take action.”

In addition to the contribution from AWS, in-kind support for the observatory includes database management and components provided by Radboud University Medical Centre; back end curation and web interface provided by the Fraunhofer Institute for Digital Medicine; AI development in identifying disease patterns supported by contextflow GmbH; and scientific and medical expertise provided by the Medical University of Vienna. Participating research institutions include hospitals in 19 countries. The project is also supported by the Republic of Korea.

“The ZODIAC Observatory has a global scope, so we rely on collaboration with and support from scientific and industry partners around the world,” said Najat Mokhtar, IAEA Deputy Director General and Head of the Department of Nuclear Sciences and Applications. “By working together, by sharing data and expertise, we can strengthen countries’ capacity to respond faster and more effectively to emerging diseases and to prevent them from developing into new pandemics.”