Invest1 publisher2 min readPublished
Official: reimbursement for Vuno's new code could become benchmark for Korea's medical AI industry
Exemptions that let Korean medical AI run conditionally in hospitals are expiring. Vuno's review is the first to finish, and the procedure code that follows carries a reimbursement rate later entrants inherit.
The Investor · Invest desk
What happened
- NECA held its final deliberation on September 18 on Vuno's VUNO Med-DeepCARS, including the function that monitors a patient's risk of cardiac arrest within 24 hours.
- Expiring exemptions push the rest of the cohort into the same queue, with Lunit, JLK and Deepnoid entering assessment sequentially from early next year.
- JLK's conditional use period for its cerebral infarction subtype classification tool JBS-01K runs out in February next year, and Lunit's INSIGHT MMG is in its final exemption extension.
- Clearing the assessment removes the temporary use period, the mandatory reporting of institutions using the product, and the written patient consent that hospitals currently have to collect.
- National health insurance coverage is decided only after the assessment is complete, so approval and payment are two separate milestones for these products.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint An industry official told the Seoul Economic Daily that the reimbursement level of the first case could become the benchmark for judging the business sustainability of the entire industry. Lunit, JLK and Deepnoid have not begun their reviews.
- cost Hospitals have to keep patient costs down to adopt these tools in volume, so between approval and a set rate the bill sits with the hospital or the patient.
- decision Winning a rate turns on outcome and cost-of-stay evidence, so the cohort's next spending choice is trials and health economics work.
- capability A procedure code is what makes insurance coverage discussions possible at all for products that could previously only be used conditionally in selected clinical settings.
If the September 18 deliberation on VUNO Med-DeepCARS ends in approval, the new medical procedure code that follows is available to companies that were never in that review [1][8]. "Later-entrant AI companies can also use this code, and on that basis, discussions on applying national health insurance coverage can begin," said an official in the medical AI industry quoted by the Seoul Economic Daily [9].
An official at Vuno said the company understands the review has been completed: "No separate findings or requests for supplementary material have been delivered to the company so far, so we are waiting for the official announcement" [2][3]. Until now these tools have run in selected clinical settings under the assessment exemption or the innovative medical technology program [5]. The coverage decision is a separate process that starts once the assessment closes [14]. The report does not include a proposed rate or a date for that decision [25].
Vuno is documenting reductions in cardiac arrest and death together with shorter stays in general wards and intensive care units [18]. Lunit is trying to show that INSIGHT MMG does more than assist existing radiology readings, improving cancer detection rates and patient outcomes [19]. Those are clinical and economic studies, and the budget going into them is not going into model work.
Samjong KPMG put South Korea's AI healthcare market at $377 million in 2023 and projected average annual growth of 50.8% [20]. At that rate the market doubles in about 1.7 years, and three years of compounding takes the 2023 base to roughly $1.29bn [21][22]. How much of that gets billed as a diagnostic or treatment procedure depends on rates set after assessments like this one [4][14].
Three outcomes are plausible. A rate generous enough to pay for continuous monitoring gives the queue behind Vuno a workable price. A low one damages Vuno and, according to a second industry official quoted by Sedaily, hurts later entrants using the same code as well [17]. The third is narrow scope: DeepCARS monitors the risk of cardiac arrest within 24 hours, while Lunit's product screens for breast cancer [1][12]. In my view the scope of that first code matters more to Lunit than its level does. A code drafted for AI-assisted analysis as a class would prove that wrong, and would make JLK's February expiry and Deepnoid's turn next year negotiations about a number already settled in Vuno's file [11][13].
What to watch
- NECA's official announcement on VUNO Med-DeepCARS and the wording of the procedure code it creates.
- Whether JLK's JBS-01K assessment, due after its February expiry, is judged against Vuno's code or a separate one.
- Whether Lunit's current exemption extension is its last before INSIGHT MMG enters assessment.