Enabling scalable AI for Digital Health: interoperability, consent and ethics support

This paper proposes an approach for building scalable AI applications in digital health, with a specific focus on addressing interoperability, consent and ethics challenges. These challenges need to be considered in the context of increasingly available tooling for streamlined model development, training, validation, and deployment, while accommodating novel solutions for explainable AI support for clinicians. Such an approach is required because digital health ecosystems involve many data type created by different systems, and often used as part of workflows over different jurisdictional boundaries. Interoperability solutions are needed to support technical and business agreements between parties providing data and services, including knowledge intensive services, such as ML and AI. Computable expression of consent and ethics policies are needed to control how patient information is used, including compliance with regulative rules, possibly from different policy contexts. Our approach, based on the latest interoperability and enterprise policy standards may provide a useful guidance for the practitioners building scalable AI solutions for digital health.