The global artificial intelligence (AI) healthcare market is predicted to grow rapidly. Various technologies for AI-based Software as Medical Device (SaMD) have been developed, and demand for their health insurance reimbursement coverage is increasing. Reimbursement policies for new medical technologies need to be thoroughly examined, despite their role in stimulating the market. The reason is that health insurance finance can have significant impact on the entire country, including patients, providers, and industry.

Based on guidelines for applying Korea’s innovative medical technologies, especially AI-based imaging medical technology, to health insurance, we examined outcome factors such as procedures and benefits. After the guidelines’ publication in 2019, we examined their impact on the medical device market through changes in the number of clinical trials and identified cases in which health insurance was listed.

The process of registering SaMD’s health insurance occurs in accordance with the existing medical technology evaluation system, and it can take up to 460 days from application to approval. If new technologies, including SaMDs demonstrate significant improvement in diagnostic capabilities and cost-effectiveness compared to existing practices, separate health insurance claims are available. Since the scheme’s announcement in 2019, items approved for SaMD clinical trials have increased (2018: ‘n=4; 2020: n=44; 2021: n=37). However, as of November 2022, only one was listed for health insurance benefits (VUNO Med-DeepBrain®), and one case was not listed on benefits but was recognized for its innovation and entered the market on the premise of suspending the health technology assessment process and accumulating real-world data (VUNO Med-DeepCARS®).™ DeepBrain® is a deep learning-based image reading technology costing about KRW80,000 (USD60) higher than conventional brain-magnetic resonance imaging and readings.

The number of SaMDs attempting clinical trials is increasing, but there is a low number of cases of reimbursement because most technologies are often classified as existing technologies and do not receive additional compensation. Since SaMD continuously is developed by accumulating data and feedback, a flexible system that can reflect this is required.