Development and depolyment of artifical intelligence-based solutions in healthcare - from artisanal manufacture to production line?

Abstract

Data-driven digital solutions can help healthcare actors with better preventive, therapeutic and welfare recommendations, contribute to the development and implementation of artificial intelligence-based decision support in diagnostics and therapy, and advance biomedical research, in particular the development of personalised therapies and diagnostics. Health artificial intelligence research predominantly focuses on technological or ethical aspects, and there are few system-level studies investigating the future integration of AI developments into digital health care models. In a reference project within the framework of Semmelweis University's Thematic Excellence Programme, the production line of the artificial intelligence algorithm is being completed. In this article, we seek to answer the question: how can we enable the parallel development of multiple AI modules over a long period of time in the right legal, regulatory and funding environment? Can we design the development and deployment process at system level as one or more production lines? The complexity of the process requires a holistic approach, as it is necessary to think about the innovative aspects of the chosen technology, the added value it brings, along the lines of the qualification and clinical trials and the technology assessment, already at the development stage. We have to plan and organise the process accordingly, and to involve the funder, healthcare providers, technology services, regulators, citizens/patients, academic researchers: the actors of the digital ecosystem. In the Data-Driven Health (D2H) Division of the National Laboratory for Health Security, which is being launched in early 2022, we aim to build on the results of this paper to (1) provide a picture of the international uptake of AI developments and applications for health, (2) summarise the risks of system-level adoption and their specifics in Hungary, and (3) attempt to identify the steps required for widespread domestic adoption.