[Game Changer] Lunit on how AI can revolutionize cancer treatment
Published: 01 Oct. 2023, 06:00
Updated: 03 Oct. 2023, 13:15
- LEE JAE-LIM
- lee.jaelim@joongang.co.kr
Artificial intelligence is penetrating every corner of people's daily lives, but Korean startup Lunit wants to put the technology to use in a more critical realm: fighting cancer.
Lunit is the first Korean company to participate in the Joe Biden administration’s cancer-focused initiative, Cancer Moonshot, which has a hefty aim of reducing the death rate from cancer by at least 50 percent over the next 25 years.
The Kosdaq-listed company is among the few unicorn startups based in Korea with its corporate valuation estimated at 2.5 trillion won ($1.85 billion). The startup's total amount of investment accumulated to 159 billion won.
“Just like self-driving cars, an era will arrive when AI will autonomously decipher and analyze medical data and detect abnormalities,” Lunit CEO Suh Beom-seok said. “When that time arrives, pathologists’ daunting workload — from analyzing CTs, X-rays, MRIs to conducting and checking biopsies — those mundane, repetitive tasks which are done by human experts, will be done by AI.”
Founded in 2013, the medical software company has come a long way. Initially an AI-powered fashion startup launched in August 2013 by six graduates of Korea Advanced Institute of Science and Technology (KAIST) under the company name “Cldi,” it used image recognition technologies to match ideal clothes or accessories.
Suh joined the company in 2016 as a chief medical officer after Cldi rebranded itself as Lunit and changed its business focus to medical AI. Suh studied medicine at Seoul National University’s College of Medicine and has a bachelor’s degree from KAIST in biological sciences.
Suh became the new boss of Lunit in 2018, tapped by previous CEO and co-founder Paek Seung-wook, to propel the company forward in the medical AI sector.
In August, Lunit announced its 10-year road map to post an operating profit of 5 trillion won and a revenue of 10 trillion won by 2033.
In the January-June period, Lunit’s revenue jumped 200 percent on year to 16.4 billion won. Operating loss was 12 billion won during the same period. Suh anticipates Lunit will turn a profit by 2025.
Lunit’s AI software is currently utilized by 2,000 medical facilities worldwide and Suh confidently forecasts the number will keep growing.
Lunit’s two flagship products are Lunit Insight and Lunit Scope. Lunit Insight is an AI software that screens X-rays to detect 10 different abnormalities in breast and lung cancer. Lunit Scope is an AI-powered biomarker that can locate the density of cancer immune cells around the cancer area from a patient’s tissue slide to predict the treatment effect of anticancer drugs.
The average accuracy of Lunit products are estimated to be between 95 to 97 percent, according to the company.
“A day will come when Lunit’s AI hits an accuracy of 99.9 percent,” Suh said. “That is where Lunit’s competitive edge lies — its precision, which will perform better than its human counterparts. This will go hand in hand with reducing medical costs for patients as well. In the case of the U.S., 20 percent of the nation’s total gross domestic product is spent to cover health care costs. As far as I know, a substantial amount is used to cover the medical fees in Korea as well. Utilizing AI will definitely bring down the cost.”
The Korea JoongAng Daily sat down with Suh for an interview at Lunit’s headquarters in Gangnam District, southern Seoul, to discuss the company’s ongoing projects and its plans for the global market.
This interview has been edited for length and clarity.
Lunit is the first Korean company participating in the U.S. government-led project called Cancer Moonshot. What’s happening there? What kinds of projects are you participating in, and how much funding are you receiving from the U.S. government?
Lunit is one of the founding members at the heart of the initiative. It is important to note that Lunit is a founding member — that’s different from other companies which only formed partnerships. However, the project is still very much at the starting line, so nothing much is happening at this point. We are gathering periodically, at least once every three months to discuss what we can do together to fulfill what the White House set out to accomplish. Once specific agendas are set, that is when I believe what Lunit can contribute to the initiative will be more clearly outlined.
Nearly 86 percent of the company’s revenue is generated overseas. Which region is Lunit seeing the most growth?
Japan. Lunit is fast growing in the region, much faster than other countries. A primary factor is the fact that Lunit’s AI-powered products and services received the green light from Japan’s Health Ministry for health care reimbursement.
What this means is that the Japanese government provides subsidies to the hospitals installing our solutions. So from the perspective of health care providers, they have no reason not to accept our AI products when they contribute to raising the facilities’ revenue and improving the health of their patients — it’s more the merrier.
It's only in Japan that hospitals are subsidized for using Lunit’s AI technologies. Domestically, Korean hospitals need to pay out of their own pockets to utilize Lunit's solutions and not all institutions have enough budget to do so. That’s really the wedge delaying nationwide distribution of our technologies.
What about in other countries? If insurance doesn’t cover the cost of deploying your technologies, how can Lunit enter other regions?
That is why Lunit’s global expansion, for now, depends on partnerships with health care solution providers like Hologic, GE HealthCare, Sectra AB, Philips, AGFA HealthCare, IBM and Guardant Health.
To push forward the drive to distribution, Lunit products need to be covered by insurance in respective nations.
Lunit is working toward removing this obstacle, and I anticipate that within the first half of 2024, our solutions will be insured within Korea as well. Once this wedge is removed, Lunit’s revenue will grow at an unprecedented rate.
We can even imagine that once our AI products outperform humans, Lunit can cover the insurance fees, but accepting way less than patients are paying now. If Lunit’s AI does reach the accuracy of 99.9 percent, the 0.01 percent possibility of error can be covered by Lunit, however slim the chances may be. That also means less pressure for the doctors as well, since they no longer have to worry about the responsibility of error falling upon them.
At the press event to celebrate Lunit’s 10-year anniversary in August, you emphasized the potential growth of Lunit Scope, even more so than Lunit Insight. Based on your earnings report from the January-June period, Lunit Insight takes up 56.68 percent of your total revenue, while Lunit Scope takes up 34.38 percent. What kind of possibilities does Lunit Scope hold?
Let me put it this way: Lunit Insight is limited in aiding what human doctors already do fairly well, whereas Lunit Scope itself can be defined as a medical practice. It does not need human assistance — it’s an AI biomarker which can locate the density of immune phenotypes near the cancer area, and predict how a certain antibody is effective against pathogens by tracking target-specific proteins, the former latching on to kill the latter.
Lunit Scope has greater potential because we can diversify the product lines to act as biomarkers for each of the new anticancer drugs pharmaceutical companies roll out. In other words, Lunit Scope and the new drugs will be a pair. Once a new drug is approved to be commercialized, all cancer patients will need to go through Lunit Scope for it to accurately predict the immunotherapy response of the new drug.
Lunit Scope’s next step is to develop new anticancer drugs. Right now, the global anticancer drug market is focused on finding cures that can eliminate the gene mutation that causes cancer. However, if the scene shifts toward immunotherapy, it will become another market opportunity for Lunit.
Cancer immunotherapy does not, and cannot directly kill the cancer cells. It treats cancer by juicing up the body’s natural immune system so that it can identify and kill those cells. With proper target identification and the right kind of partnerships, Lunit Scope can be utilized to develop new drugs that can modulate immune responses to treat cancer.
Of course, we cannot achieve all of this alone. We are considering partnerships, mergers and acquisitions to meet our objective and the companies are not necessarily limited to Korea.
Lunit’s next 10-year road map also includes developing a big data platform that can be universally utilized and deployed across all medical institutions across the globe. What is this about?
Our vision is to create a universal platform from which all hospitals can freely upload and download information related to cancer. But the thing about AI is that no matter how much the software is meticulously trained, it functions differently based on the types of data that is fed into the system.
The problem here is that data changes constantly. It evolves. In order to keep up with the data flow, AI needs to be constantly connected to the new data streams so its functions can be updated without cessation.
That is why we’ve come to a conclusion that we need this platform. It is the only way we can ensure our products’ precision. We have several options to create this platform. We are considering partnerships with top vendors of electric medical records (EMR) software, and companies with picture archiving and communication systems (PACS) to gain the database. Another way would be to develop an in-house platform with a specialty in one area that hospitals find irresistible, making it difficult from them not to utilize our platform.
Again, we are not planning on doing this alone. We are [also] deliberating on acquisitions, if necessary, to accomplish this, but everything is up in the air because, well, this is part of our 10-year road map and we have another decade ahead of us.
BY LEE JAE-LIM [lee.jaelim@joongang.co.kr]
with the Korea JoongAng Daily
To write comments, please log in to one of the accounts.
Standards Board Policy (0/250자)