[Column] Medical device approval: AI could be the answer – with a caveat

Korea’s medical device industry ranks seventh worldwide in terms of capabilities, and aims to achieve an even higher ranking in the future.

According to data from the Korea Health Industry Development Institute in 2023, Italy and the United Kingdom rank first and second, with approximately $16 billion, which is nearly twice as high as Korea's $7 billion. However, considering the growth potential, Korea could achieve the same level within a few years if the current growth rate is maintained.

The various medical devices we use in the field are more closely related to patient safety and treatment effectiveness than to industrial aspects, as they are considered public goods. Therefore, industrial growth is directly linked to the quality of healthcare and the lives of citizens.

However, this industry, which had been growing at an average annual rate of 8 percent over the past few years, has recently slowed down. While quantitative growth has been achieved, the industry is experiencing typical growing pains, with qualitative growth lagging behind.

To overcome this problem, efforts have been made to harmonize international regulations on various medical devices, which form the foundation of the global market. Regulations were also established in the field of artificial intelligence, attracting global attention as the first of its kind worldwide.

However, as with all policies, there are gains and losses. The heightened regulatory barriers have caused manufacturers to hesitate in developing new products, and the prolonged review periods have led to fatigue among both the industry and those involved in review and certification processes.

The core issue lies in the approval process. As safety standards for medical devices have increased, approval requirements have undergone comprehensive upgrades, resulting in the creation of a high-cost regulatory structure.

Testing and certifying organizations have generated significant revenue through this process, which is a double-edged sword. On the positive side, Korea's expertise and infrastructure in medical devices have reached world-class levels. However, on the negative side, this imposes substantial cost burdens on manufacturers.

An even bigger problem is objectivity. No matter how clear the guidelines are, subjective judgment inevitably comes into play when humans conduct reviews. What if similar cases result in different outcomes depending on the reviewer? This is extremely frustrating for manufacturers.

Korea already has a well-established legal framework for medical devices, including the Medical Device Act, various ordinances, and guidelines. All review materials are documented and organized in a format that is easy for AI to learn, which can be viewed as an optimal environment for AI adoption.

If an AI-based review system is introduced, it could reduce the current approval process, which takes over six months, to within three months and achieve results comparable to those of the U.S. Food and Drug Administration's “ELSA (Enterprise Language Support Assistant)” system.

The second is cost reduction. If AI is introduced to perform tasks that currently require skilled reviewers, such as verifying requirements and reviewing draft reviews, the reduction in labor input will allow for maintaining a higher review standard while significantly reducing review fees. This could lead to additional benefits, such as reinvesting the saved costs into supporting innovative products.

The third is enhancing objectivity. Since human error is inevitable, maintaining consistency has been a chronic issue, often leading to disputes. However, AI can ensure objective and fair reviews based on consistent standards, thereby improving customer satisfaction.

The fourth is predictability. Manufacturers can design products with licensing in mind from the development stage, reducing the time and costs required for pre-licensing preparations. This ultimately motivates the release of diverse products.

In the current regulatory environment, where valuable administrative resources are consumed in responding to various complaints and 70 percent of cases require additional measures due to differences in institutional perspectives, the first step toward creating an ecosystem that fosters productive cycles for medical device development lies in adopting AI for responding to and reviewing various complaints.

The increasing number of faster and more diverse medical devices entering the market ultimately means broader treatment options for patients. If innovative devices developed by Korean SMEs can be commercialized quickly, the competitiveness of our medical field will also improve accordingly.

Of course, AI is not a panacea. Complex and new forms of medical devices will still require expert judgment, and the reliability and transparency of AI systems themselves must be ensured.

However, change is already underway. Overseas, cases of regulatory innovation utilizing AI are emerging one after another. We, too, must now turn technological possibilities into reality.

If there are concerns that human roles will disappear, this is not the case. In a reality where regulation is perceived as consumption, the transition of roles will establish a foundation for higher-quality growth by integrating regulators and reviewers to support patient safety and recovery.