Applications, challenges and a trustworthy use of artificial intelligence in public health

The rapid advancements in artificial intelligence (AI) over recent years have resulted in its integration into people's everyday lives. The wide availability of diverse data within the public health sector opens up a number of fields of application for AI, ranging from infection research and the analysis of epidemiological data to the extraction of information from communication data such as social media, the development of new resilience strategies against climate change and the systematic evaluation of specialist literature. The quality of the underlying data is paramount to the successful implementation of AI applications. In public health research, on the one hand, there is a wide variability of data types including, but not limited to, image data, numerical data and survey data. On the other hand, availability can be limited, for example when a rare pathology is being investigated and/or stringent data protection requirements apply. Concurrently, it is imperative to maintain high ethical standards and to mitigate biases, imbalances and lack of transparency as early as possible. We delineate an approach towards the responsible and trustworthy utilisation of AI applications in public health, which leads from the initial question to the data and the model development to evaluation and emphasises the importance of careful and complete documentation.