OBJECTIVES
- To identify use cases, where AI/ML have been applied:
- To establish or assist diagnosis
- To establish an optimal treatment pathway or treatment dosing
- In other healthcare applications
- To understand regulatory requirements and the scope of evidence that need to be generated for using AI/ML in the healthcare setting.
METHODS A targeted narrative literature review was performed. Scientific publications (identified in PubMed) and grey literature (identified on the websites of regulatory agencies FDA and EMA) were reviewed and analyzed. RESULTS Several cases have been identified. For example, HeartFlow® is an AI based deep learning technology for non-invasive diagnosis and management of coronary artery disease based on a computerized tomography coronary angiogram results and patient characteristics, implemented across NHSs hospitals. HeartFlow® is backed up by a prospective pragmatic trial and peer-reviewed studies addressing improvement in clinical decision making and patients’ quality of life and reduction in the cost of care. Other identified applications pertain to decision support for intensive care patient’s discharge (Pacmed), management of chronic diseases (various providers), pharmacotherapy decision support (Pacmed), prognosis prediction (Pacmed), image processing (various providers), and compliance monitoring (AiCure®). From the regulatory perspective, AI/ML based solutions can be considered medical devices (Software as Medical Device, SaMD). Regulatory requirements depend on the level of risk associated with the use of the device. EMA imposes stricter regulatory requirements than the FDA. CONCLUSIONS The number of promising AI/ML based technologies is increasing, but few have been implemented widely at the point of care. The need for external validation, data exchange and privacy, and logistics of implementation remain the main obstacles.
