Korea rewrites how digital therapeutics must prove they work

AI-generated illustrationAI-generated illustration
South Korea’s Ministry of Food and Drug Safety published a revised guideline on 10 September 2026 covering how clinical trial protocols for digital therapeutic devices must be written. The notice comes from the ministry’s Medical AI Software Division and sits under the country’s Digital Medical Products Act.
A digital therapeutic is software prescribed as treatment — an app that delivers cognitive behavioural therapy for insomnia, or a program that retrains attention. Korea approved its first in 2023 and has been building the evidentiary machinery since. The hard problem is not whether the software runs; it is what counts as proof that it worked, and how much improvement is enough to matter to the patient rather than to the statistics.
According to Financial News, the revision adds concrete efficacy endpoint criteria and minimum clinically important difference standards, with worked examples for multi-centre and follow-up trials, across seven indications: nicotine and alcohol use disorder, panic disorder, depression, eating disorder, ADHD, and mild cognitive impairment. The ministry’s own notice confirms the revision and its date; the detailed contents sit in attached documents rather than on the page.
Regulators setting endpoints decide, in practice, which claims a piece of software is allowed to make about a mind.