In an age where emphasis on the use of artificial intelligence (AI) for the good of humanity is increasing, last week’s AI for Good Global Summit at the International Telecommunication Union (ITU) brought leading experts together to demonstrate the multidimensionality of AI in emergency response and outbreaks, as well as in risk reduction.

“Can we predict disease outbreaks using public health data, health records, population density, income levels as well as weather, wind speed, previous outbreaks and location proximity?” asked Dominic Haazen, lead health policy specialist, World Bank​, and panel moderator at the AI for Good Global Summit.

From the use of hyper-targeted advertising for public health messaging to the use of big data for providing information during disasters and outbreaks of epidemics, to using images, convolutional neural networks and video sequencing for detecting malnutrition in children, and the use of social media analyses to track health trends, experts are optimistic about the use of artificial intelligence in emergency response and outbreaks.

For instance, hyper-targeted advertising helps to send messages to a target population rather than a one-size-fits-all message, thereby making it possible to determine which health message would work best for a particular population using machine learning, said Ingmar Weber, research director for social computing, Qatar Computing Research Institute. Machine learning will help generate personalised health messages based on different features of a target population, namely age, gender, marital status, recent parent, education, nationality, weight loss, or even geo-fencing, which can for instance target people that have recently been to the hospital, he added.

According to Jeanine Vos, head of SDG Accelerator, GSMA, mobile big data can be used for providing information during outbreaks or disasters while protecting and respecting the privacy of individuals, for instance, information about the movement of people during an outbreak of an epidemic. It can also help government and emergency agencies to better understand, predict, plan and respond to emergencies, she said.

Clara Palau Montava, technology team lead, UNICEF, added that big data is used for estimating risk factors. She further demonstrated that the use of messaging platforms through SMS to collect information in real time was vital to the Ebola crisis response in 2015, as seeing statistics in real time made it possible to respond rapidly in real time. For instance, people were asked through messages on how to prevent Ebola and where to get help, she said.

However, she also pointed out that big data can be unrepresentative, for instance, when those who do not have phones are taken into consideration, and added that unrepresentative data would lead to unrepresentative outcomes. She further mentioned that there is a team of five data scientists working with other broader networks of collaborators to tackle this problem through data science.

When asked about the greatest challenge about the widespread adoption of these innovations, Montava highlighted that access to readable data is challenging, either because it is not in a format that can be read or because it was inconsistently gathered. She also pointed out that data collaboration and partnership with private sectors as well as actors in the field would be helpful to build upon the work of others. Other experts also mentioned the challenge of explaining AI, which is difficult for people to understand, and the importance of investing in capability skills and resources for those in the technical field, government bodies, NGOs, and agencies.