The concept of modern medicine as we know it, centering on evidence-based diagnoses, was formed only a couple of decades ago and has since become the basis of today’s medical textbooks, theory and practice.
Although modern medicine is effective, it has room to improve by incorporating AI technology, said Anthony Paek, co-founder and executive chairman of South Korean medical AI solution startup Lunit.
Anthony Paek, co-founder and chairman of Lunit, gives a presentation at an AI forum held on Oct. 25 in Seoul.
“Data-driven medicine using AI technology has some similarity with evidence-based medicine as both analyze data to detect illnesses, but the difference is that deep-learning technology performs better than humans,” said Paek during a presentation at an AI forum organized by SK Research Institute, a think tank run by SK Group, Oct. 25 in Seoul.
“I believe AI technology will take medicine to the next level,” he added.
In some research projects, Lunit’s AI detection solutions for breast and lung cancers have outpaced radiologists in detecting cancers and improving the cancer detection rate by doctors, according to the Lunit chairman, who earned a doctorate in computer vision at KAIST, a leading science and tech university in Korea.
The lung cancer detection rate by the company’s software in one of the projects, for example, reached nearly 99.8 percent while the figure achieved by radiologists came in at some 90 percent.
The data used in the project was provided by Korea’s four largest hospitals: Samsung Medical Center, Asan Medical Center, Seoul National University Hospital and Yonsei Severance Hospital.
Although there is significant untapped potential in the medical industry, the biggest hurdle for medical startups such as Lunit is that it takes too long to complete clinical trials, a problem that tends to deter potential investors.
Developing a solution that can roll out a high cancer detection rate regardless of variables such as the use of different medical instruments at different hospitals or differences in breast density among trial subjects, is another challenge for Lunit.
Despite such challenges, the company’s AI solution has so far turned in a better performance than those developed by European and American companies, Park said.
The outcome of a comparative study of Lunit’s solution and other AI-based cancer detection solutions will be announced in December at the annual meeting of the Radiological Society of North America, a leading medical research conference for radiology and medical professionals.
Founded in 2014, the Korean startup with 100 employees and researchers has so far received some 30 billion won ($25.6 million) in funding. Some 30 percent of its workforce consists of AI experts who have presented at least one published report at a top-tier AI conference.
By Kim Young-won (firstname.lastname@example.org)