On September 6, China's National Medical Products Administration granted conditional approval to isotretinavir hydrochloride tablets (brand name Aipushiwei) for treating mild to moderate COVID-19 infection in adults. The drug was developed by Westlake University, Westlake Laboratory and Westlake Pharmaceuticals (Hangzhou) Co., Ltd., led by Professor Huang Jing of Westlake University's School of Life Sciences and Wang Tingliang, an assistant to the university president. According to the university, this is the first original drug in China to be approved for market after AI-assisted development, taking three and a half years from initial discovery to the completion of clinical trials.
The drug has also been classified as a Category 1 innovative drug, the top tier in China's drug registration classification, which requires an entirely new chemical structure and an entirely new mechanism of action.
Where AI came into the process
The starting point was a DNA-encoded compound library containing 49 billion compounds. The persistent problem with libraries of this kind is screening: there are simply too many molecules, and even high-throughput experiments cannot keep up — a single misconfigured condition can wash out an effective molecule entirely. The team's approach was to have an AI model help determine screening conditions and control compounds, then pick out the batch most likely to be genuinely effective from the sequencing data.
The funnel narrowed sharply. The pool of 49 billion was cut down to just over 100 candidate molecules, then further down to 9 that entered synthesis and testing, of which 6 showed strong inhibitory activity. The compound that was ultimately advanced was designated WLU6937. The university describes the process as a closed loop: computational prediction, experimental validation, data fed back into the model, and the model refined again — it took several rounds before the conditions stabilized.
DEL technology itself was first proposed in 1992, and has since remained stuck on the same bottleneck: libraries that are enormous but hard to screen accurately. Aipushiwei is being described as the world's first small-molecule innovative drug approved via DEL technology — a claim that currently comes from the developer and media reports, and has not yet been corroborated by any third-party statistical source.
What conditional approval actually means
Conditional approval does not mean the full approval process has been completed. Under China's current rules, this type of approval is granted on the basis that early clinical data has already shown benefit, and the company is required to submit confirmatory study data after the drug reaches market — failure to do so can result in the drug being pulled from the market. Public reports have not disclosed the specific enrollment numbers, primary endpoints, or control design behind this approval, nor has there been any word on pricing or the timeline for inclusion in the national medical insurance list.
For users in China, the near-term, perceptible changes are limited. The oral COVID-19 drug market was crowded around 2022, and current demand is nowhere near that scale, so the drug's actual sales volume cannot yet be assessed. Its more concrete significance lies in the method: a pipeline that puts AI into the early-stage hit-compound discovery step has, for the first time, been run end to end in China — from library to approval document. The three-and-a-half-year figure sets a reference point that future projects in the queue can now be measured against.
Sources: National Medical Products Administration approval information, Westlake University public statements, multiple financial media reports, CocoLoop; compound library size, number of candidate molecules and development timeline verified against the university's own account.