NaLaA – Corona Global

Short Description

Event-based surveillance (EBS) is a key component in public health surveillance. Its effectiveness is conditioned by the number of sources used and the speed of processing its information. Current Artificial Intelligence (AI) methodology allows the filtering and extraction of key information from online texts with high precision, much higher than current keyword-based approaches. The advantage of these algorithms is that they use textual context, which is crucial in a health-based setting where details beyond the loose key words determine the significance of detected events. Together with Africa CDC, we will use expert-labeled data to train such novel AI algorithms for EBS and also label new sources such as social media. These algorithms help prioritize the most important news trained on prior extraction work done by Africa CDC and based on jointly labeled data. Afterwards, we will adopt used sources and algorithm to the needs of Africa CDC. To ensure sustainability of the project, we set clear performance goals for each step which measure that a sufficient amount of the code is documented and tested and that its performance is in accordance with what Africa CDC needs for their daily EBS-based screening. Documentation and testing will also enable us to open-source our project and share the idea with more users, ensuring they can easily utilize our work with minimal friction. In parallel, we will conduct workshops with our partners and together with ProMED conduct training on-site to improve EBS efforts beyond the use of algorithms.

Project Objectives

Event-Based Surveillance (EBS) is a fundamental pillar in Africa CDC’s infectious disease surveillance routine and its continent-wide mandate. Thus, the implementation of novel AI methodology for EBS promises the most impactful increase in value.

Our base goal is the exchange of knowledge in surveillance and AI with our colleagues from Africa CDC. This exchange should evolve in the development of a software tool that aims to improve signal detection from news articles and social media.

Most importantly, we try to achieve sustainability in this project by laying the foundation for a prolonged exchange of expertise among both institutes. Furthermore, we try to ensure the persistent advancement of the project software by heavily involving our colleagues from Africa CDC in the software development process.

Last update: June 2021