Low-Code/No-Code Artificial Intelligence Platforms for the Health Informatics Domain

Abstract

In the contemporary health informatics space, Artificial Intelligence (AI) has become a necessity for the extraction of actionable knowledge in a timely manner. Low-code/No-Code (LCNC) AI Platforms enable domain experts to leverage the value that AI has to offer by lowering the technical skills overhead. We develop domain-specific, service-orientated platforms in the context of two subdomains of health informatics. We address in this work the core principles and the architectures of these platforms whose functionality we are constantly extending. Our work conforms to best practices with respect to the integration and interoperability of external services and provides process orchestration in a LCNC modeldriven fashion. We chose the CINCO product DIME and a bespoke tool developed in CINCO Cloud to serve as the underlying infrastructure for our LCNC platforms which address the requirements from our two application domains; public health and biomedical research. In the context of public health, an environment for building AI driven web applications for the automated evaluation of Web-based Health Information (WBHI). With respect to biomedical research, an AI driven workflow environment for the computational analysis of highly-plexed tissue images. We extended both underlying application stacks to support the various AI service functionality needed to address the requirements of the two application domains. The two case studies presented outline the methodology of developing these platforms through co-design with experts in the respective domains. Moving forward we anticipate we will increasingly re-use components which will reduce the development overhead for extending our existing platforms or developing new applications in similar domains.