Weekly
Implantable speech BCIs have mostly been tested with participants repeating words chosen by researchers. This participant picked words, answered open questions and used the device to call a grandchild.
The hardware remains investigational. Paradromics plans six years of follow-up on safety and durability, so routine use is not on the table.
09-14Zhejiang covers one low-cost BCI service, leaving invasive surgery and consumables self-paid. Class II device review drops to 40 working days; 2025 industry revenue exceeds 2 billion yuan. Research hospitals receive funding and waived bed quotas, testing clinical-research incentives.
09-17First same-human comparison of an endovascular stent electrode and scalp EEG. The stent electrode gives stronger per-channel modulation and better blink rejection, but adds ECG crosstalk; left-right limb resolution remains unresolved. It shows the signal gain from closer-to-cortex recording.
09-15Only 20% of 129 studies share code or pipelines, explaining why cross-dataset training so often loses accuracy.
09-16Preprocessing gets its own control track, and 25 times more external data bought limited gains.
09-18Waitlist-controlled crossover in children with acquired brain injury, testing EEG markers and whether training beats standard care.
09-15Coordination across surgery, rehab and imaging is the bottleneck; this hospital writes MDT into its trial protocol.
09-14Beijing's 7,980-yuan price tops all provinces while Class A reimbursement covers four hospitals; delivery rests on how private markets value local firms.
09-14Jiangxi caps mutual ethics review at 10 working days, a rare measurable clock for multicentre trials that other provinces can copy.
