UN Global Pulse worked with the World Health Organization (WHO) to explore the use of data from radio talk shows to signal early warnings of health risks and health-related matters.
Population movements, behavioural changes, food production, and many other factors linked to globalisation and economic development are responsible for the continuous emergence of infectious hazards. Diseases such as SARS, avian influenza, and COVID-19 represent new challenges for outbreak alert and response.
The term epidemic intelligence refers to activities related to early identification of potential health hazards, their verification, assessment, and investigation in order to recommend public health control measures. The World Health Organization started an initiative related to these topics called EIOS (Epidemic Intelligence from Open Sources). EIOS is a network of organizations with one common goal: to minimise the impact of emerging risks to human health through quality and timely detection, assessment, early warning, forecasting, and reporting. In doing so, the initiative aims to help strengthen the capacity of countries with respect to early detection and evidence-based decision making.
UN Global Pulse joined forces with EIOS to explore the utility of mining radio broadcasts for early signals of health risks. The initial use case analysed data from radio broadcasts in Uganda over a period of 18 months using the speech-to-text technology that Pulse Lab Kampala developed to transcribe radio talk into text. The study investigated key health-related signals based on specific keywords provided by the EIOS team.
Preliminary results showed potential correlations between health topics people discussed on the radio and recorded health metrics on the ground. The results were presented and discussed during the EIOS Global Technical Meeting that took place in November in Seoul, South Korea. UN Global Pulse and EIOS will work to verify these early signals in order to develop a model that can be integrated into the initiative’s applications.