WHO and UN Global Pulse, the Secretary-General’s digital innovation initiative are embarking on a new venture to analyse content shared on public radio across the African continent to support infodemic monitoring. Part of the wider infodemic management programme, this collaboration will integrate radio data analysis into monitoring tools to inform the infodemic response by including more voices of vulnerable populations.
Radio remains one of the most reliable and affordable ways of accessing and sharing information with some 44,000 stations worldwide as reported by UNESCO. People discuss everything from the light-hearted to the serious, including the effects COVID-19 is having on them, their families, and their communities. Often, opinions expressed in radio talk shows include first-hand accounts of incidents, and mis- and disinformation reported by citizens that may not be recorded elsewhere.
Dr Sylvie Briand, Director WHO Global Infectious Hazards Department said, “Listening to and understanding the concerns and questions of communities is pivotal in every health emergency to help us shape our risk communication interventions. This tool will be a key addition to our infodemic toolbox.”
UN Global Pulse has worked on various projects to convert radio speech into text for vernacular languages and to analyse the information to understand perceptions. This collaboration will expand that scope by developing methods to capture and analyse content on topics that discuss COVID-19 and vaccination efforts across Africa.
Miguel Luengo-Oroz, Chief Data Scientist at UN Global Pulse added, “As we develop new methodologies and deploy new tools, we want to give public health officials the most useful insights and create new scientific evidence so the field of infodemic management continues to build solid foundations.”
The project is already off the ground with pilot studies to analyse information related to vaccines. In the coming months, the plan is to integrate more than 100 public radio stations and develop a custom-built monitoring and analysis dashboard that will allow public health officials, infodemic managers, and fact checking organizations to track changing questions and concerns, including mis- and dis-information, to gather insights that can inform public health decisions.
The methodology will account for potential algorithmic biases along the AI analysis pipeline, and the project will implement data privacy instruments developed by UN Global Pulse to mitigate the potential risks and harms associated with this technology.
By incorporating languages and voices that may often be excluded from official reporting and analysis, this effort aims to de-marginalise communities by better understanding their changing needs in public health emergencies.
In future, custom language models could be built to incorporate a wider variety of languages and dialects spoken, and support routine social listening of radio narratives as a way to continue strengthening accountability to affected populations for global health.