04/11/2026
Can AI choose acupuncture points as well as an experienced practitioner?
A new study published this October in JMIR Formative Research put this to the test. Researchers gave real clinical cases to ChatGPT and DeepSeek, in both English and Chinese, and asked them to build treatment protocols. Ten senior TCM physicians then rated them blind, alongside the original protocols written by physicians.
What they found:
🔹 On some criteria, the AI protocols scored as high as the physicians' protocols, or even higher.
🔹 Language mattered: DeepSeek performed better in Chinese, ChatGPT in English.
🔹 Where AI still falls short: individualized treatment, and consistency across languages.
To me, the most interesting part isn't who "won." It's what the study confirms: point selection is knowledge that can be organized, structured and made accessible. The personal judgment of an experienced practitioner, though, still can't be replaced.
That's exactly the balance we believe in at AcuRing. We make the knowledge of points accessible through technology, and keep it complementary to the practitioner's work, never a replacement for it.
Practitioners: would you trust AI to suggest points to you? Or to your patients?
🔗 Liu X, Guo Z. JMIR Formative Research, Oct 2026. DOI: 10.2196/95851