Airdoc-AIFUNDUS met its retinopathy targets in China's review

What happened

Shanghai Airdoc Medical Technology built Airdoc-AIFUNDUS, software that applies deep learning based on convolutional neural networks to fundus photographs of adults with diabetes. It advises doctors whether moderate non-proliferative diabetic retinopathy or worse is present and whether the patient needs further examination. The Center for Medical Device Evaluation (CMDE) of China's National Medical Products Administration reviewed the company's application for Class III registration as an innovative device. The trial was the maker's own, a prospective, multicenter, single-arm study with senior doctors' reading as the reference standard. The report says that sensitivity and specificity “达到预期目标值” (reached the preset targets) and that no adverse event occurred. It gives no figures and names neither the sites nor the number of patients. The report adds that “与预期受益相比较，综合剩余风险可接受” (compared with the expected benefit, the residual risk is acceptable). On 31 July 2020, CMDE recommended approval. It asked the company to study the accuracy of referrals further after approval, including use in primary care. The report does not say that the software's advice guided any patient's care.